MerQuR+Statistics Glossary

Statistics Glossary

Conceptual explanations of the analyses in MerQur and their parameters: what an analysis is for, how it works and how each option changes the result. Each entry first gives an everyday example and then the actual result MerQur gives on independent validation data.

#ABCDEFGHIJKLMNOPQRSTUVWYZ

#
  • ε-neighbourhood and minimum points (DBSCAN)conceptIn DBSCAN, the radius of the neighbourhood around a point (ε) and the minimum number of points (minPts) that this neighbourhood must contain for the point to count as the core of a dense region.
A
  • Abbott correction (natural response)conceptA method used in dose-response experiments to correct the response rate due to treatment by accounting for responses that occur even without treatment (such as natural deaths in the control group).
  • Absolute agreement and consistency (ICC types)conceptTwo definitions of the intraclass correlation coefficient: one measuring whether raters give the same values (absolute agreement), the other whether they merely rank subjects in the same order with the same spacing (consistency).
  • Accelerated failure time (AFT) model and survival distributionsconceptA parametric survival model in which covariates stretch or shrink the time to the event by a multiplicative factor, with log(time) described by a linear model and a chosen error distribution (Weibull, log-normal, log-logistic and others).
  • Adaptive Gauss-Hermite quadrature (AGQ) and Laplace approximationconceptMethods that numerically compute the likelihood integral over the random effects in generalized linear mixed models, which has no closed-form solution.
  • Age Standardization and SMRanalysisA method that makes disease or death rates of populations with different age (and optionally sex) structures comparable by removing the effect of the age difference.
  • Alternative-specific constants (ASC)conceptConstant terms in choice models such as conditional logit, defined relative to a base option, that capture each option's average appeal not explained by the measured attributes.
  • Analysis of variance (ANOVA) tableconceptA summary table that partitions the total variability into its sources (model, factors, residual) and gives, for each source, the sum of squares, degrees of freedom, mean square, F and p value.
  • ANCOVAanalysisAnalysis of covariance: comparing group means on an outcome while statistically holding constant the effect of one or more continuous variables (covariates) that also influence the outcome.
  • Anderson-Darling testconceptA goodness-of-fit test of whether a sample comes from a specified distribution (usually the normal), based on a comparison of cumulative distributions that gives special weight to the tails.
  • AnisotropyconceptDirection dependence in how quickly a spatial variable loses similarity with distance; for example, values remaining related over long distances in one direction and only over short distances in another.
  • Anomaly DetectionanalysisMethods that use statistical or machine-learning criteria to identify and flag unusual (outlying or anomalous) observations that do not fit the general structure of a data set.
  • ARCH-LM test (conditional heteroscedasticity)conceptA Lagrange multiplier test of whether, in the residuals of a time series model, today's volatility depends on the volatility of past periods (an ARCH effect).
  • Area under the ROC curve (AUC)conceptThe probability that a classifier or diagnostic test gives a randomly chosen positive case a higher score than a randomly chosen negative case; a summary of discriminating power across all thresholds.
  • ARIMA ForecastanalysisA method that forecasts the future values of a time series with an autoregressive integrated moving average (ARIMA) model, which draws on the series' own past values and past forecast errors.
  • ARIMA orders (p, d, q) and driftconceptThe three numbers that define the structure of an ARIMA model: the number of autoregressive terms (p), the number of differences (d) and the number of moving-average terms (q); drift is a constant term in the differenced series.
  • Asymptotic (normal) approximationconceptApproximating the sampling distribution of a test statistic or estimate by the normal (or chi-square) distribution that holds in large samples.
  • Autocorrelation (ACF) and partial autocorrelation (PACF) functionsconceptFunctions showing a time series' correlation with its own past values at different lags (ACF), and the direct correlation remaining once the effect of the intermediate lags has been removed (PACF).
  • Autocorrelation correction and prewhitening (Hamed-Rao, TFPW)conceptMethods used with the Mann-Kendall trend test that stop positive autocorrelation in a series from misleading the test, either by correcting the variance of the test statistic or by removing the autocorrelation from the series.
  • Auxiliary variables (imputation model)conceptVariables that are not part of the main analysis but are used as predictors in the multiple imputation of missing data because they are related to the incomplete variable or to the reason it is missing.
  • Average marginal effect (AME)conceptIn nonlinear models (logit, probit, Tobit), the effect of a change in a predictor on the probability or expected value of the outcome, averaged over all observations in the sample.
  • Average treatment effect (ATE, ATT, ATC)conceptThree versions of the average causal effect of an intervention, depending on the population it is defined for: the whole sample (ATE), the treated (ATT) and the untreated (ATC).
B
  • Bandwidth (kernel smoothing)conceptIn smoothing methods such as kernel density estimation and local regression, the parameter that sets how far away observations can be and still contribute to the estimate at a point, and hence how smooth the result is.
  • Bartlett's test of equal variancesconceptA chi-square test of whether two or more groups have equal variances, assuming the data are normally distributed.
  • Bayes factor (BF₁₀)conceptA ratio showing how many times more probable the observed data are under one hypothesis (H₁) than under the other (H₀); it measures the relative strength of evidence between two hypotheses.
  • Bayesian ANOVAanalysisAn analysis of variance that judges whether group means differ by comparing models with and without the factors through Bayes factors, giving the strength of evidence for each effect.
  • Bayesian CorrelationanalysisAn analysis that uses a Bayes factor to assess whether the relationship between two variables differs from zero, and the posterior distribution of the correlation coefficient to show how strong it is.
  • Bayesian Linear RegressionanalysisA method that estimates a linear regression explaining an outcome by several predictors, combining prior information about the coefficients with the data and giving a posterior distribution and credible interval for each coefficient.
  • Bayesian t-Test (JZS Bayes Factor)analysisA test of whether a mean, or the difference between two means, is zero, using a Bayes factor that shows how strongly the data support the 'difference' and 'no difference' hypotheses relative to each other.
  • BCa bootstrap confidence intervalconceptA bias-corrected and accelerated confidence interval that shifts the percentile limits to account for bias and skewness in the bootstrap distribution.
  • Best linear unbiased prediction (BLUP)conceptIn mixed models, the prediction of each group's random effect (a school, patient or plant family, for instance) obtained by shrinking the group's data toward the overall mean.
  • Between-study variance (τ²) and heterogeneity (I²)conceptTwo measures in meta-analysis of how much the true effect sizes of the combined studies differ: τ² is the absolute size of that difference, and I² is the share of the observed spread that is due to it.
  • Between-subjects factorconceptA factor in which each subject appears at only one level, so that its levels separate different groups of subjects (for example sex or treatment group).
  • Binomial TestanalysisA test that checks exactly, using the binomial distribution, whether the success rate observed in a two-outcome variable (yes/no, success/failure) is consistent with a pre-specified proportion.
  • BiplotconceptA chart that shows observations (or row categories) and variables (or column categories) together in the same two-dimensional plane.
  • Bland-Altman Agreement AnalysisanalysisAn analysis that assesses whether two methods measuring the same thing can be used interchangeably, by computing the mean difference between them (bias) and the limits of agreement within which most individual differences fall.
  • Blocking (block variable)conceptA design principle that groups experimental units into similar sets (blocks) according to a characteristic that affects the outcome but is not of interest, and compares treatments within each block, removing that characteristic's variability from the error term.
  • Bonferroni correctionconceptA correction that keeps the probability of at least one false significant result below α when m tests are run, by judging each test at α/m (or multiplying the p-values by m).
  • Bootstrap (resampling)conceptA method that draws thousands of new samples with replacement from the sample at hand to compute the uncertainty of a statistic directly.
  • Bootstrap CIanalysisA method that finds the confidence interval of a statistic without distributional assumptions or a ready-made formula, by drawing thousands of new samples with replacement from the data at hand and recomputing the statistic in each.
  • Bootstrap likelihood ratio test (BLRT)conceptA test in latent class or latent profile analysis of whether a K-class model fits significantly better than a K - 1 class model, with the distribution of the likelihood ratio generated by parametric bootstrap.
  • Bootstrap optimism correctionconceptAn internal validation method that uses the bootstrap to estimate how optimistic a prediction model's performance on its own development data is, and subtracts that optimism.
  • Box plotconceptA chart that summarizes a variable's median, quartiles, spread and outlying values with a single box and two whiskers.
  • Box's M test (equality of covariance matrices)conceptA multivariate test of whether the covariance matrices of several groups are equal.
  • Box-Cox transformationconceptA family of power transformations, governed by a parameter λ, that aims to make the variance of a positive variable more constant and its distribution more symmetric.
  • Box-Tidwell test (linearity of the logit)conceptA test, in logistic regression, of the assumption that a continuous predictor is linearly related to the logit (log-odds).
  • Breslow-Day test (homogeneity of odds ratios)conceptA test, for 2 x 2 tables split by the strata of a confounding variable, of whether the odds ratio is the same in every stratum.
  • Breusch-Pagan and White testsconceptTests of whether the variance of regression residuals changes with the explanatory variables (heteroscedasticity).
  • Brown-Forsythe F* testconceptAn F test of the equality of means that can replace one-way ANOVA when the group variances are not equal.
  • Brunner-Munzel testconceptA rank test for two independent groups of whether the probability that a randomly chosen value from one group is smaller than one from the other differs from 0.5, without requiring equal variances or equal distribution shapes.
C
  • Canonical CorrelationanalysisA multivariate analysis that summarises the relationship between two sets of numeric variables by finding a weighted combination from each set that correlates as strongly as possible with the other.
  • Cause-specific hazardconceptIn survival data with competing risks, the instantaneous rate of a particular type of event among people who have not yet experienced any event.
  • Censoring (survival analysis)conceptNot knowing exactly when the event happened for an observation; most often because the event has not yet occurred when the study ends.
  • Change-point tests (Pettitt, Buishand, SNHT)conceptTests that determine whether the mean of a time series shifts abruptly at an unknown time and, if so, when that shift most likely occurred.
  • Chi-Square Goodness of FitanalysisA test of whether the observed distribution of a single categorical variable fits a pre-specified distribution (equal proportions, theoretical ratios, population percentages).
  • Chi-Square IndependenceanalysisA test of whether two categorical variables are associated, comparing the observed counts in a crosstab with the counts expected if there were no association.
  • Choosing the number of clustersconceptDeciding how many groups to split the data into in clustering or latent class analysis, guided by indicators such as the elbow, silhouette, gap statistic or information criteria.
  • Chow structural break testconceptAn F test of whether a regression relationship is the same before and after a pre-specified point (e.g. a date).
  • Clark-Evans nearest-neighbour indexconceptA ratio that compares the average distance from points to their nearest neighbours with the distance expected under a completely random pattern, indicating whether the pattern is clustered, random or regular.
  • Class imbalance and balanced weightsconceptThe situation in a classification problem where one class has far fewer observations than the others, and the inversely proportional weights given to classes to stop this from pushing the model toward the majority class.
  • Classification table (confusion matrix)conceptA table that cross-tabulates the classes predicted by a model against the true classes and shows the kinds of correct and incorrect classifications.
  • Classification threshold (cut-off)conceptThe boundary that decides from which value a model's predicted probability or score is treated as the 'positive' class.
  • Cluster-robust (clustered) standard errorsconceptSandwich-type standard errors that allow observations within the same cluster (school, patient, firm) to be dependent.
  • Cluster-Robust Cox (Marginal and Frailty Models)analysisA method for data with several event times from the same person, family or centre, which estimates Cox regression while allowing for the dependence between these observations, either correcting the standard errors at the cluster level or modelling the differences between clusters with a frailty term.
  • Cochran's Q TestanalysisA test of whether the success (yes) rates of three or more yes/no conditions measured on the same people are equal; it generalises the McNemar test to more than two conditions.
  • Cochran-Mantel-Haenszel (CMH) TestanalysisA test of the association between two binary variables that combines the 2x2 tables at each level of a third, confounding variable, keeping that variable's influence out of the comparison.
  • Coded factors (response surface)conceptIn response-surface and experimental-design analyses, factors converted from their original units to a common scale, usually running from -1 to +1.
  • Coefficient of determination (R²) and adjusted R²conceptThe proportion of the total variability in the dependent variable that is explained by the model; adjusted R² penalizes that proportion for the number of variables in the model.
  • Coefficient of variation (CV)conceptThe ratio of the standard deviation to the mean; a unit-free measure that expresses spread as a percentage of the mean.
  • Coefficient path and ridge traceconceptA plot, used in regularized regression, of how each coefficient changes as the penalty parameter is varied over a grid of values.
  • Cohen's d (effect size)conceptAn effect size that expresses the difference between two means in standard deviation units.
  • Cohen's d for paired data (d_z, d_av, d_rm)conceptEffect sizes that standardize the mean difference between two measurements on the same people, differing in which standard deviation goes in the denominator.
  • Cohen's hconceptAn effect size for the difference between two proportions, expressed on the scale of the arcsine-transformed proportions.
  • Cohen's Kappa (including Fleiss κ and Krippendorff α)analysisA coefficient that measures how well two or more raters agree when sorting the same items into categories, after removing the agreement that would occur by chance alone.
  • Cointegration (Johansen / Engle-Granger)analysisA method (Johansen, Engle-Granger) that tests whether two or more series, each non-stationary and drifting freely over time, move together in the long run, that is, whether there is an equilibrium relationship between them.
  • Column proportions z-testconceptA post-hoc method that compares, within each row of a crosstab, the percentages of the column groups pair by pair with a two-proportion z test and marks those that differ with letters.
  • Common-language effect size (CLES)conceptAn effect size expressed as the probability that a randomly chosen person from one group has a higher value than a randomly chosen person from the other group.
  • CommunalityconceptThe proportion of a variable's variance that is explained by the common factors extracted in factor analysis (or by the components retained in principal component analysis); written h².
  • Compact letter display (homogeneous subsets)conceptA method that turns pairwise comparison results into a short display in which groups that do not differ significantly share a common letter.
  • Competing RisksanalysisA survival analysis for situations with several event types, where the occurrence of one makes the others impossible, that estimates the cumulative probability of each event over time and the effects of factors on these events.
  • Complete and quasi-complete separationconceptIn logistic-type models, a situation in which a predictor or combination of predictors separates the outcome perfectly or almost perfectly, so the maximum likelihood estimates run off to infinity.
  • Complete spatial randomness (CSR)conceptThe null hypothesis, defined by the homogeneous Poisson process, that points are scattered over the study area independently of one another and with the same intensity everywhere.
  • Conditional Logit (Choice Model)analysisA model for situations in which a person chooses one of several alternatives, estimating how the alternatives' features (price, time, quality) affect the probability of being chosen by comparing within each choice set.
  • Conditional variance (volatility) and GARCH ordersconceptThe way a series' uncertainty for the next period changes over time with past shocks and past volatility; the quantity that GARCH models describe.
  • Confidence interval for a proportion (Wilson, Clopper-Pearson, Agresti-Coull, Jeffreys)conceptAn interval expressing the uncertainty about the population value of a proportion observed in a sample (e.g. 12 out of 40), and the various methods for constructing it.
  • Confidence level and confidence intervalconceptAn interval around an estimate that covers the true value with a given long-run success rate; the confidence level is that rate (90%, 95%, 99%).
  • Confirmatory Factor Analysis (CFA)analysisAn analysis that tests whether the items of a scale load on latent factors specified in advance from theory as expected, i.e. whether a proposed measurement model fits the data.
  • Confounding variableconceptA third variable that is related both to the exposure or group under study and to the outcome, and so distorts the relationship between them.
  • Confusion Matrix MetricsanalysisAn analysis that computes performance measures such as accuracy, sensitivity, specificity, predictive values, F1, MCC and kappa from the confusion matrix, which cross-tabulates a classifier's or diagnostic test's predictions against the true classes.
  • Conjugate prior (normal-inverse-gamma, Zellner's g)conceptA prior distribution that, combined with the likelihood, yields a posterior from the same family as the prior, so that the posterior can be computed in closed form.
  • Contingency coefficient CconceptA measure derived from the chi-square statistic that expresses the strength of association between two categorical variables on a scale from 0 to an upper bound below 1: C = √(χ² / (χ² + n)).
  • Continuity correction (Yates)conceptA correction that shifts a discrete (count-based) test statistic half a unit toward its expected value when approximating it with a continuous distribution, to reduce the error caused by discreteness.
  • Contrast and custom contrast (L)conceptA weighted linear combination of group means or model coefficients, chosen to answer the research question directly, together with a test of whether that combination equals zero.
  • Convergence and maximum iterationsconceptIterative algorithms stop when the change between successive steps falls below a tolerance (convergence), and an upper limit on iterations keeps them from running forever.
  • Correlation AnalysisanalysisAn analysis showing how two numeric variables vary together, giving the direction (positive or negative) and strength of the relationship in a single coefficient between −1 and 1.
  • Correspondence Analysis (CA / MCA)analysisA method that visually summarises the association in a crosstab of two (or more) categorical variables by placing the row and column categories on a low-dimensional map according to their similarity.
  • Cost parameter C and the ε-insensitive bandconceptThe C parameter of a support vector machine, which balances the width of the margin against the penalty for training errors, and ε, which sets the unpenalized error band in support vector regression.
  • Covariance (correlation) structure (UN, VC, AR(1), exchangeable)conceptThe model assumption that specifies which pattern the variances and correlations among observations from the same unit follow in repeated or clustered measurements.
  • Covariance and correlation matrixconceptA square table holding the variances of several variables on its diagonal and the covariance (or correlation) of each pair of variables off the diagonal.
  • CovariateconceptA variable, usually continuous, that is not of primary interest but is added to the model because it affects the dependent variable, so that its influence is statistically controlled.
  • Cox model residuals (martingale, deviance, Cox-Snell, dfbeta)conceptTypes of residuals used to examine the fit of the Cox proportional hazards model, the functional form of covariates and the influence of individual observations on the results.
  • Cox Regression (Proportional Hazards Model)analysisA survival model that shows how several factors together affect the time until an event (death, relapse, failure, leaving a job), giving a hazard ratio (HR) for each factor.
  • Cramér's V and Cohen's wconceptEffect sizes derived from the chi-square statistic that show the strength of the association between two categorical variables (or of the gap between observed and expected distributions).
  • Credible interval and highest density interval (HDI)conceptIn Bayesian analysis, an interval that contains a parameter with a given probability (for example 95%) according to its posterior distribution; the HDI is the narrowest such interval.
  • Cronbach's Alpha and Reliability AnalysisanalysisAn analysis that quantifies how consistently the items brought together to measure the same characteristic are answered, that is, the internal consistency reliability of a scale.
  • Cross-correlation function (CCF)conceptA function that computes the correlation between two time series when one is shifted relative to the other by different lags, showing which series moves ahead of the other and by how many periods.
  • Cross-loadingsconceptLoadings that show how an indicator (item) relates to factors or constructs other than its own.
  • Cross-sectional dependence (Pesaran CD)conceptIn panel data, correlation between the residuals of different units (countries, firms, regions) in the same period; Pesaran's CD test checks whether such correlation is present.
  • Crossed Mixed Model (Crossed LMM, Crossed Random Effects Mixed Model)analysisA mixed model for designs in which two factors are crossed (every worker uses every machine, every judge scores every presentation), splitting the variability in the outcome into the two factors, their interaction and error, and assessing the reliability of the measurement.
  • CrosstabanalysisA table showing, in counts and percentages, how two categorical variables are distributed together, summarising the association between them with tests and measures of association.
  • Cumulative hazard and log-minus-log plotconceptIn survival analysis, the function showing the risk of the event accumulated up to a given time, H(t) = -ln S(t); the log-minus-log plot draws ln(-ln S(t)) against ln t to check the proportional hazards assumption visually.
  • Cumulative incidence function (CIF) and Gray's testconceptIn the presence of competing events, the function giving the probability that a particular type of event has occurred by time t, and the test that compares this function between groups.
  • Cut point (dichotomization)conceptA threshold value that splits a continuous or many-valued variable into two groups, those below and those above it.
D
  • D'Agostino's K² testconceptA normality test that assesses departure from the normal distribution by combining sample skewness and kurtosis.
  • Damped trendconceptA model component in exponential smoothing that makes the trend fade as the forecast horizon lengthens, so that the forecasts eventually level off.
  • DBSCAN / HDBSCANanalysisA density-based clustering method that treats regions where points are densely packed as clusters and points left in sparse regions as noise (outliers), without needing the number of clusters in advance.
  • DBSCAN Spatial ClusteringanalysisA clustering method that groups location points by density, marking sufficiently crowded areas as clusters and points left in sparse areas as noise, without needing the number of clusters in advance.
  • Decision Tree (CART)analysisA classification and regression method that splits the data in two again and again with 'yes/no' questions, ending in subgroups that each give a single prediction, and presents the result as easily read decision rules.
  • Degrees of freedomconceptThe number of independent pieces of information that are free to vary when a statistic is computed; it determines which theoretical distribution the test statistic is compared with.
  • DeLong test and AUC confidence intervalconceptA method that computes the standard error of the area under the ROC curve (AUC) nonparametrically and compares two AUCs measured on the same subjects.
  • Delta methodconceptA way of approximating the standard error of a function of estimated parameters by using a first-order Taylor expansion of that function.
  • Dendrogram and cophenetic correlationconceptA tree diagram showing in which order, and at what distance, observations (or variables) are merged in hierarchical clustering; the cophenetic correlation measures how faithfully this tree reflects the original distances.
  • Descriptive StatisticsanalysisThe first step of any analysis: summarising where a variable's values lie, how spread out they are and what shape they take, using a few summary numbers, tables and graphs.
  • Design effect (DEFF)conceptIn a complex sample design (clusters, strata, weights), the ratio of an estimate's variance to the variance expected from a simple random sample of the same size.
  • Deterministic terms (constant, trend)conceptNon-random components added to time series models that are known functions of time, such as a constant and a linear trend.
  • Deviation (effect) codingconceptA coding scheme that represents each level of a categorical variable as its deviation from the grand mean, constraining the level effects to sum to zero.
  • Difference-in-Differences (DiD)analysisA method that estimates the causal effect of an intervention by comparing the before-after change in outcome in a group exposed to the intervention with the change in a group that was not.
  • DifferencingconceptA transformation that subtracts from each value of a time series the previous value (or the value one season earlier) in order to remove trend and non-stationarity.
  • Differential item functioning (DIF)conceptThe situation in which members of different groups (e.g. women and men) who stand at the same level of the trait being measured have different probabilities of answering or endorsing an item correctly.
  • Direct and indirect standardization (SMR)conceptTwo methods that make the rates of populations differing in a stratifying variable such as age structure comparable, by removing the effect of that difference.
  • Discriminant Analysis (LDA / QDA)analysisA multivariate method that finds the combinations of continuous variables that best separate groups known in advance, and uses these combinations to estimate which group a new observation belongs to.
  • Distance measure (Euclidean, Manhattan, Gower, great-circle)conceptThe definition that turns how dissimilar two observations are into a number, used in clustering, scaling and spatial analyses.
  • Distribution family (GLM)conceptThe choice in a generalized linear model that sets the conditional distribution of the response and therefore the relationship between its mean and variance.
  • Distribution-free (order-statistic) confidence intervalconceptA confidence interval for quantities such as the median or other percentiles that sorts the data and uses observations at particular ranks as its limits, requiring no distributional assumption such as normality.
  • Domain (subpopulation) estimationconceptEstimating a mean, proportion or total for a subpopulation (domain), such as women or a particular region, from survey data collected with a complex sample design.
  • Doubly robust estimationconceptAn approach that estimates a treatment effect in observational data using both a propensity score (treatment assignment) model and an outcome model, and stays consistent as long as either of the two is correct.
  • Dummy (indicator) variable codingconceptA way of entering a categorical variable with k levels into a regression model as k - 1 zero/one variables, each representing one level.
  • Dunn's testconceptA post hoc method used after the Kruskal-Wallis test that compares group pairs with a z test on the mean ranks obtained by ranking all the data together.
  • Dunnett's test (comparison with a control)conceptA multiple-comparison test that compares several treatment groups with a single control group only, while controlling the familywise error rate across those comparisons.
E
  • Early exaggeration (t-SNE)conceptA t-SNE setting that multiplies the high-dimensional similarity probabilities by a factor during the first phase of optimization, so that clusters separate early in the low-dimensional map.
  • Early stoppingconceptA regularization technique for iteratively trained models (gradient boosting, neural networks) that prevents overfitting by ending training when performance on a held-out validation set stops improving.
  • Edge (boundary) correctionconceptCorrections in the analysis of spatial point data that remove the bias arising because some neighbours of points near the boundary of the study area lie outside it.
  • Effect sizeconceptA measure of how large a difference or relationship is, independent of sample size.
  • Effect Size CalculatoranalysisA tool that computes not just whether a finding is statistically significant but how large it is, using standard, comparable measures (such as Cohen's d, η², Cramér's V, r and the odds ratio).
  • Effective dose (ED50, LD50) and dose-responseconceptIn a dose-response relationship, the dose estimated to produce a response in a given proportion (e.g. 50%) of subjects; when the response is death it is called the lethal dose (LD).
  • Egger's and Begg's testsconceptTests in meta-analysis of whether the funnel plot is asymmetric, i.e. whether small studies systematically report different effects from large ones.
  • Eigenvalue and eigenvectorconceptThe vectors that a square matrix (e.g. a correlation or covariance matrix) only rescales without changing their direction (eigenvectors), and the rescaling factors themselves (eigenvalues).
  • Elastic net and the L1 ratioconceptA regularization method that shrinks regression coefficients with a mixture of the L1 (lasso) and L2 (ridge) penalties, with the share of each set by the L1 ratio.
  • Elbow method and k-distance plotconceptA visual method for choosing clustering parameters that looks for the 'elbow', the point where the curve of a criterion plotted against the parameter clearly bends.
  • EM (expectation-maximization) algorithmconceptAn iterative algorithm that finds maximum likelihood estimates in models with unobserved (latent or missing) variables by alternating an expectation step and a maximization step.
  • Engle-Granger two-step methodconceptA method that tests whether two or more unit-root series move together in the long run (cointegration) by checking whether the residuals of a regression between their levels are stationary.
  • Epsilon squared (ε²)conceptAn effect size that estimates the proportion of variance explained by a factor while correcting the upward bias of η².
  • Equivalence test (TOST)conceptA test used to show that the difference between two means lies within a pre-specified range regarded as negligible, i.e. that the groups are practically equivalent.
  • Error bars (mean ± CI plot)conceptVertical lines drawn around each group mean in a chart to show the uncertainty of the estimate or the spread of the data.
  • Error distribution (normal, Student-t)conceptThe probability distribution assumed for a model's random error (innovation) term; it determines the likelihood, the tail behaviour and the forecast intervals.
  • Estimated marginal means (EMM)conceptMeans for each level of a factor computed from the fitted model, with the other factors balanced and covariates held at a fixed value (usually their mean).
  • Eta squared (η²) and partial eta squared (η²ₚ)conceptEffect size measures in ANOVA that give the proportion of variance explained by an effect.
  • Event study designconceptAn extension of difference-in-differences that estimates the effect of an intervention separately for each period before and after it, showing how the effect evolves over time and revealing any pre-trends.
  • Exact test and exact p-valueconceptA test that computes the p value from the exact null distribution of the test statistic rather than from an approximating distribution (normal, chi-square).
  • Exogenous variables (VARX, ARIMAX)conceptExternal variables in a time-series model that the model itself does not explain but that are assumed to influence the series being explained.
  • Expected frequencyconceptThe number of observations a table cell would be expected to contain if the null hypothesis were true; chi-square tests compare the observed counts with these values.
  • Expected mean squares (EMS)conceptThe expected value of each mean square in an ANOVA table, expressed in terms of variance components and fixed effects; it determines the correct F denominator and the variance component estimates.
  • Exploratory Factor Analysis (EFA)analysisA method that tries to explain the relationships among answers to many items by a few common factors that lie behind them and cannot be measured directly, showing which item belongs to which factor.
  • Exponential Smoothing (Holt-Winters)analysisA family of methods (SES, Holt, Holt-Winters) that produce short-term forecasts by continually updating a series' level, trend and seasonal pattern with weighted averages that give more weight to the recent past.
  • Exponential smoothing parameters (α, β, γ, φ)conceptThe parameters in exponential smoothing methods that set how quickly the level, trend and seasonal component of a series are updated with new observations, and how strongly the trend is damped.
F
  • F testconceptA test based on the ratio of two variance estimates (mean squares) that follows an F distribution under the null hypothesis.
  • F1 score, MCC and balanced accuracyconceptThree classification performance measures computed from the confusion matrix, used to avoid the misleading picture plain accuracy gives when classes are imbalanced.
  • Factor (component) loadingsconceptCoefficients showing how strongly an observed variable (item) is related to a factor or component; what the factors represent is interpreted from these loadings.
  • Factor and component scoresconceptScores giving each observation's value on the latent factors found by a factor analysis, or on the components of a principal component analysis.
  • Factor extraction method (ML, PAF, minres)conceptThe method that sets the criterion by which factor loadings are estimated from the observed correlation matrix in exploratory factor analysis.
  • Factor rotation (varimax, promax, oblimin)conceptA transformation in factor analysis or principal components that turns the factor axes into an easier-to-interpret 'simple structure' without changing the fit of the model.
  • Fail-safe NconceptThe number of additional unpublished studies with zero effect that would have to be added to make a significant pooled meta-analytic result non-significant.
  • False discovery rate (FDR, Benjamini-Hochberg, Storey q)conceptThe expected proportion of false positives among the results declared significant; a more lenient control criterion than the familywise error rate when many tests are run.
  • Fine-Gray subdistribution hazard modelconceptA regression model for survival data with competing risks that models directly the effect of covariates on the cumulative incidence of the event of interest.
  • Finite population correction (FPC)conceptA correction that shrinks the variance of estimates by the factor (1 − n/N) when, under sampling without replacement, the sample is a sizeable part of the population.
  • Firth correction (penalized likelihood)conceptAn estimation method for logistic regression that adds a small penalty to the likelihood to reduce small-sample bias and to give finite coefficients even under complete separation.
  • Fisher classification functions and canonical discriminant functionsconceptIn discriminant analysis, the linear score functions computed separately for each class to assign observations, and the linear components that best separate the groups.
  • Fisher z transformation (correlation confidence interval)conceptA transformation that moves a correlation coefficient onto an approximately normal scale with constant variance, used for confidence intervals and comparisons.
  • Fisher's Exact TestanalysisA test of whether two categorical variables are associated that uses exact probabilities instead of the chi-square approximation, especially for 2x2 tables with small samples.
  • Fisher's exact testconceptA test of independence between two categorical variables in a contingency table that computes the exact probability of the observed table, or a more extreme one, with the row and column totals held fixed.
  • Fisher's LSD (unadjusted pairwise comparisons)conceptA post-hoc method that compares pairs of groups after ANOVA with t tests using the pooled error variance, without any multiple-comparison adjustment.
  • Fixed and random effectsconceptThe distinction between modelling a factor's levels as all the levels of interest (fixed) or as a sample drawn from a wider population (random).
  • Fleiss' κ and Krippendorff's α (multiple raters)conceptCoefficients that measure how well more than two raters agree when classifying or scoring the same units, correcting for chance agreement.
  • Folded F test (variance ratio)conceptA test of whether two groups have equal variances, based on the ratio of the larger variance to the smaller one.
  • Forecast error variance decomposition (FEVD)conceptA decomposition in a VAR model showing what share of the error variance of a variable's h-step-ahead forecast comes from its own shocks and what share from shocks to the other variables.
  • Forecast horizon and forecast intervalconceptHow many periods beyond the last observation a time series model forecasts (the horizon), and the interval around each forecast within which the future value is expected to fall with a given probability.
  • Frailty modelconceptAn approach in survival analysis that models the unobserved risk shared by observations in the same cluster (family, hospital, recurrent events within a person) through a random effect that multiplies the hazard function.
  • Frequency weight (summary-table data)conceptA weight stating how many identical observations a data row stands for; used when data are entered as a count table rather than as individual records.
  • FriedmananalysisA test that needs no normality, for designs in which the same people (or blocks) are measured under three or more conditions: it checks whether the conditions differ systematically, based on how each person ranks the conditions within themselves.
  • Full information maximum likelihood (FIML)conceptAn estimation method for data with missing values in which each case contributes to the likelihood only through its observed variables, so incomplete cases are not discarded.
  • Funnel plotconceptA plot of each study's effect size against its precision (usually its standard error), used in meta-analysis to look visually for small-study effects such as publication bias.
G
  • G-test (likelihood-ratio chi-square, G²)conceptThe likelihood-ratio counterpart of Pearson's chi-square test, comparing observed and expected frequencies through their log ratios.
  • Games-Howell testconceptA post hoc test that compares groups pairwise when variances and group sizes are unequal.
  • Gap statisticconceptA criterion in cluster analysis for choosing the number of clusters by comparing within-cluster spread for different numbers of clusters with what would be expected in cluster-free (uniform) reference data.
  • GARCH (Volatility Model)analysisA conditional variance model that describes and forecasts how the volatility (variance) of a return series changes over time, taking into account the clustering of calm and turbulent periods.
  • GEE working correlation and marginal modelsconceptAn approach for repeated or clustered data that approximates within-cluster dependence with a working correlation matrix and estimates population-averaged (marginal) effects.
  • Generalized Additive Model (GAM, Nonlinear Regression)analysisA regression method that models the outcome as a sum of smooth curves, one per explanatory variable, whose shapes are learned from the data rather than specified in advance, so that nonlinear relations are captured flexibly.
  • Generalized cross-validation (GCV)conceptA criterion that gives a fast approximation to leave-one-out cross-validation for linear smoothers and penalized regression, used to choose the smoothing or penalty parameter.
  • Generalized Estimating Equations (GEE)analysisA method for dependent data, such as repeated measurements on the same person or observations within the same cluster, that estimates population-averaged regression coefficients for continuous, binary or count outcomes with standard errors that are robust to the dependence.
  • Generalized eta squared (η²G)conceptA proportion-of-variance effect size designed to make effects comparable across between-subjects, within-subject and mixed designs.
  • Generalized Linear Mixed Models (GLMM)analysisA regression method for binary or count outcomes in data where observations are grouped in clusters such as patients, schools, herds or sites, modelling them with random effects that vary between clusters in addition to fixed effects.
  • Generalized linear model (GLM)conceptA modelling framework that extends linear regression to non-normal outcomes (binary, counts, positive continuous) through a link function and a distribution family.
  • Geometric and harmonic meanconceptTwo kinds of average used instead of the arithmetic mean for positive data that grow multiplicatively or are expressed as rates.
  • Getis-Ord Gi* HotspotanalysisA local clustering statistic that shows, with a z-score for each location, whether the sum of values at that location and its neighbours is significantly higher (hot spot) or lower (cold spot) than expected across the area.
  • Glass's ΔconceptA standardized effect size that divides the difference between two group means by the standard deviation of the control group only.
  • Goodness-of-fit tests (Pearson X² and deviance)conceptTests of how well the frequencies or probabilities predicted by a model match the observed data; the most common are the Pearson X² and deviance (G²) statistics.
  • Gradient BoostinganalysisA machine learning method that builds a strong predictive model by adding small decision trees one after another, each new tree correcting a little more of the errors left by the previous ones.
  • Granger causalityconceptThe idea, and test, of whether the past values of one time series help predict the future values of another beyond what that series' own past already provides.
  • Greenhouse-Geisser and Huynh-Feldt correctionsconceptMethods that correct the p value in repeated measures ANOVA when sphericity is violated, by shrinking the F test's degrees of freedom by a factor ε.
  • Grubbs' and generalized ESD testsconceptTests of whether the observation farthest from the mean (Grubbs) or up to r observations (generalized ESD) are outliers, in data assumed to be normally distributed.
  • GWR (Geographically Weighted Regression)analysisGeographically weighted regression: instead of assuming a relationship is the same across the whole area, it fits local regressions at each location that give more weight to nearby observations, showing how the coefficients change from place to place.
H
  • Haldane-Anscombe (zero-cell) correctionconceptAdding a small constant (usually 0.5) to every cell of a contingency table that contains a zero count, so that the odds ratio and its logarithm can be computed.
  • Hausman testconceptA test of a model assumption that compares an estimator that is consistent and efficient under the null hypothesis with one that is consistent in any case but less efficient.
  • Hazard ratio (HR)conceptA ratio showing how many times a factor multiplies the risk of the event occurring at any moment.
  • Hedges' gconceptA standardized mean difference that removes the small-sample upward bias of Cohen's d by means of a correction factor.
  • Heritability (h²)conceptThe proportion of the total (phenotypic) variance in a trait that is due to genetic differences.
  • HeteroscedasticityconceptA situation in a regression or ANOVA model where the variance of the errors (residuals) changes with the predictors or with the fitted value.
  • Hexbin MapanalysisA spatial summary map that divides the study area into equal-sized hexagonal cells and colours each cell by the number of points it contains (or by the sum, mean or median of a value).
  • Hidden layer and activation functionconceptThe intermediate layers of units in a neural network, between input and output, that create new features by passing weighted sums of the inputs through a nonlinear function, and the function those units use.
  • Hierarchical (blockwise) regressionconceptA regression approach in which predictors are entered in blocks in a theoretically chosen order and the added contribution of each block over the previous ones is tested.
  • Hierarchical Bayesian RegressionanalysisA multilevel regression model for grouped data, such as pupils within schools or patients within hospitals, that estimates the effects of predictors while accounting for differences between groups and gives each group an estimate improved by 'borrowing strength'.
  • Hierarchical ClusteringanalysisAn exploratory grouping method that merges observations step by step, starting with the most similar, to build a tree (dendrogram), and obtains clusters by cutting this tree at the desired level.
  • Histogram and bin countconceptA chart that splits the range of a numerical variable into equal-width intervals (bins) and shows the number of observations in each as bars.
  • Hodges-Lehmann estimator (pseudomedian)conceptA robust, rank-based estimate of location or location shift that matches the Wilcoxon and Mann-Whitney tests.
  • Holm (step-down Bonferroni) methodconceptA multiple-comparison method that sorts p values from smallest to largest and compares each with a progressively less strict threshold, controlling the familywise error rate.
  • Homogeneity of regression slopesconceptThe ANCOVA assumption that the relationship (slope) between the covariate and the dependent variable is the same in every group.
  • Homogeneity of variance assumptionconceptThe assumption that the groups being compared have equal population variances (spreads); the classical t-test and ANOVA rely on it.
  • Hosmer-Lemeshow goodness-of-fit testconceptA fit test for logistic (and probit) regression that groups observations by their predicted probabilities and compares observed and expected numbers of events in each group.
  • Hotelling's T²conceptThe multivariate generalization of the t test; a statistic that jointly tests whether a mean vector equals a given vector, or whether two groups' mean vectors are equal.
  • Hyperparameter tuning (grid search)conceptChoosing the settings of a machine-learning model that are not learned from the data (such as tree depth or a penalty parameter) from a set of candidate combinations, based on cross-validated performance.
I
  • Impulse response function (IRF) and Cholesky decompositionconceptIn a vector autoregression (VAR), the function showing how a one-time shock to one variable affects all variables in the system in later periods, and the method used to separate the shocks from each other.
  • Incidence rate ratio (IRR)conceptIn count models, the factor by which the expected number of events per unit of time or exposure is multiplied when a predictor increases by one unit.
  • Independence of irrelevant alternatives (IIA)conceptThe assumption in multinomial and conditional logit models that the odds of choosing one option over another do not depend on which other options are available.
  • Independent-Samples t-TestanalysisA test of whether the means of two independent groups truly differ, by asking whether the observed difference can be explained by chance.
  • Indirect effect (mediation)conceptThe part of the effect of a variable (X) on an outcome (Y) that passes through one or more mediators (M); in the simple model it is measured as the product of the a and b paths (a·b).
  • Inertia, singular values and profilesconceptIn correspondence analysis: the measure of the total amount of association in a contingency table (inertia), the singular values that share this amount out across dimensions, and the percentage distributions of the rows or columns (profiles).
  • Influence diagnostics (Cook's distance, leverage, DFBETAS)conceptDiagnostic statistics that measure how much individual observations change the estimates of a regression model.
  • Information criteria (AIC, AICc, BIC)conceptCriteria for comparing models that weigh their fit to the data (likelihood) against their complexity (number of parameters); the model with the smaller value is preferred.
  • Inner weighting scheme (PLS-SEM)conceptThe choice, in the PLS path modelling algorithm, of the rule by which each latent construct's inner proxy is formed from its neighbouring constructs.
  • Instrumental variable and endogeneityconceptAn approach for consistently estimating a causal effect when a predictor is correlated with the error term (endogenous), using a variable that affects that predictor but influences the outcome only through it.
  • Instrumental Variables Regression (2SLS)analysisA regression method that, when an explanatory variable is correlated with the error term (endogenous), consistently estimates its effect on the outcome by using an instrumental variable that affects the explanatory variable but influences the outcome only through it.
  • Interaction effectconceptA situation in which the effect of one variable on the outcome changes with the level of another variable.
  • Interaction plotconceptA plot that shows the means of the outcome across the levels of one factor, with a separate line for each level of a second factor, to visualise whether the two factors interact.
  • Interquartile range (IQR) and Tukey fencesconceptThe width of the middle half of the data (IQR = Q3 − Q1) and the Tukey fences that use this width to draw an outlier boundary.
  • Interval-Censored SurvivalanalysisA method for data in which the exact time of the event is unknown and it is known only to have happened somewhere between two check-ups, which estimates the survival curve and compares groups or factors.
  • Intraclass correlation (ICC)conceptThe share of total variance due to differences between groups (or between measured units), used both for measurement reliability and for multilevel data.
  • Intraclass Correlation Coefficient (ICC)analysisAn analysis that expresses, as a reliability coefficient between 0 and 1, how closely several raters, devices or measurement occasions agree when they measure the same subjects.
  • Inverse distance weighting (IDW)conceptA method that estimates the value at an unknown location by weighting the values at surrounding known points by the inverse of a power of their distance.
  • Inverse Mills ratioconceptThe ratio of the standard normal density to the cumulative distribution, λ(z) = φ(z) / Φ(z); the term added to a regression in the Heckman selection model to correct for bias caused by sample selection.
  • Inverse probability weighting (IPW)conceptA method for observational studies that weights each individual by the inverse of their probability of receiving the treatment they actually received (the propensity score), so that treatment groups are balanced on the observed confounders.
  • Isolation forest and contamination rateconceptAn anomaly detection method that finds outlying observations by how easily random splits can separate them from the rest, and the rate that sets how many observations are labelled anomalies.
  • Item difficulty, discrimination and guessing (1PL, 2PL, 3PL)conceptIn item response theory, the parameters that determine the probability that people of different ability levels answer an item correctly.
  • Item fit (infit, outfit, S-X²)conceptStatistics in item response theory and Rasch models that assess how well the observed responses to an item agree with the responses the model predicts.
  • Item Response Theory (IRT / Rasch)analysisA family of models that ties the probability of each response to the relation between person and item, estimating each item's difficulty and discrimination and each person's level on the measured trait on one common scale.
J
  • JackknifeconceptA resampling method that recomputes an estimate leaving out one observation (or one cluster) at a time and uses those replicates to estimate the standard error and bias.
  • Jarque-Bera testconceptA normality test that jointly measures how far the skewness and kurtosis of the data depart from the normal values (0 and 3).
  • Johansen trace and maximum-eigenvalue testsconceptMaximum likelihood tests, within the vector error correction model framework, of how many cointegrating (long-run equilibrium) relations link several non-stationary time series.
  • Johnson-Neyman technique (separate slopes)conceptA method that finds the range of values of a continuous moderator (covariate) over which an effect or group difference is statistically significant.
  • Jonckheere-Terpstra test (ordered alternative)conceptA rank-based test of whether the values of three or more independent groups with a natural order increase or decrease monotonically along that order.
  • JZS (Cauchy) prior and the scale parameter rconceptThe default family of prior distributions for effect size used when computing Bayes factors in Bayesian t-tests, ANOVA, correlation and regression, named after Jeffreys, Zellner and Siow.
K
  • k-fold cross-validationconceptA method that estimates how well a model will predict new data by splitting the data into k equal parts and repeatedly using one part for testing and the rest for training.
  • K-Means ClusteringanalysisAn exploratory clustering method that divides observations into a pre-set number (K) of groups so that each observation is as close as possible to the centre of its own cluster.
  • k-medoids (PAM)conceptA clustering method that uses an actual observation in the cluster (the medoid) rather than the mean as each cluster's centre and builds the clusters by minimizing total distance.
  • k-Nearest Neighbors (KNN)analysisAn assumption-free method that assigns a new observation to the majority class of the k observations whose measurements most resemble it (or predicts its numeric value as the average of these neighbours).
  • Kaiser criterion (eigenvalue > 1)conceptThe rule, in principal component or factor analysis, of retaining the components (factors) whose eigenvalue in the correlation matrix exceeds 1.
  • Kaplan-Meier estimator (survival curve)conceptA nonparametric method that estimates the survival function step by step from censored time-to-event data by multiplying the probabilities of surviving each event time.
  • Kaplan-Meier Survival AnalysisanalysisA method that summarises the time until an event (death, relapse, failure, readmission) with a survival curve, properly accounting for people who have not yet had the event during follow-up, and compares the curves of groups.
  • Kappa statistic (Cohen's κ)conceptA coefficient that measures agreement between two raters or two classifications on categorical decisions, after removing the agreement expected by chance.
  • KDE Density MapanalysisA kernel density estimation map that builds a continuous density surface from point locations, by adding up smooth 'bumps' spread around each point, to show where the points are concentrated.
  • Kendall's tauconceptA coefficient from -1 to +1 that measures the ordinal association between two variables by comparing the numbers of concordant and discordant pairs of observations.
  • Kendall's W (coefficient of concordance)conceptA coefficient between 0 and 1 that shows how far several raters agree with one another when ranking the same set of objects.
  • Kernel (weighting) function (Gaussian, bisquare, Epanechnikov)conceptA function that sets how the weight given to other observations decreases as their distance from a point grows; the basic building block of local estimation.
  • Kernel density estimation (KDE)conceptA nonparametric method that estimates the probability density of a variable, or of point locations, as a smooth curve or surface by placing a small smooth 'bump' (kernel) on each observation and adding them up.
  • KMO measure of sampling adequacy and Bartlett's test of sphericityconceptTwo preliminary checks of whether a correlation matrix is suitable for factor analysis: the Kaiser-Meyer-Olkin measure and Bartlett's test of sphericity.
  • Knapp-Hartung (HKSJ) adjustmentconceptAn adjustment in random-effects meta-analysis that rescales the standard error of the pooled effect and bases tests on the t distribution instead of the normal.
  • Kolmogorov-Smirnov test and Lilliefors correctionconceptA goodness-of-fit test that compares a sample's cumulative distribution with a theoretical distribution (or two samples with each other) through their largest vertical distance.
  • KPSS stationarity testconceptA time-series test whose null hypothesis is that the series is stationary (around a level or a trend) and whose alternative is a unit root.
  • Kriging (simple, ordinary, universal, block)conceptA spatial interpolation method that predicts the value at unsampled locations as a weighted average of nearby measurements, with weights based on the spatial relationship described by the variogram, and that also gives the uncertainty of the prediction.
  • Kruskal-WallisanalysisA test that needs no normality: it ranks the values of three or more independent groups together and compares the groups' mean ranks, testing whether at least one group differs systematically from the others.
  • Kullback-Leibler (KL) divergenceconceptA non-negative measure of divergence that quantifies how much information is lost when one probability distribution is approximated by another.
  • KurtosisconceptA shape measure based on the fourth standardized moment that shows how heavy or light a distribution's tails are compared with the normal distribution.
L
  • Lack-of-fit testconceptAn F test of whether a model describes the data adequately compared with a full (saturated) model that gives every distinct combination of predictor values its own mean.
  • Lag length selectionconceptDeciding how many past periods (lags) to include in autoregressive models (AR, VAR, the ADF unit-root test), usually with the help of information criteria.
  • Lasso RegressionanalysisA regression method that shrinks the coefficients with a penalty, setting some of them exactly to zero, and thereby selects the predictors needed for prediction from many candidates.
  • Latent Class Analysis (LCA)analysisA model that estimates how many unobservable subgroups (latent classes) lie behind the patterns of answers to categorical questions (yes/no, present/absent and so on), and the probability that each individual belongs to each group.
  • Latent Profile Analysis (LPA)analysisA method that uses a mixture model to find unobservable subgroups (latent profiles) sharing similar patterns of values on continuous measurements, and estimates the probability that each individual belongs to each profile.
  • Learning rate (shrinkage)conceptA coefficient that sets how large the update made at each step of an iterative learning algorithm is.
  • Leave-one-out (LOO)conceptA method that drops one observation (or one study) at a time and refits the model on the rest, to assess predictive performance or how sensitive the results are to a single unit.
  • Levene's testconceptA test of whether two or more groups have equal variances (spreads).
  • Life table (actuarial method)conceptA method that divides survival times into predefined intervals and computes the probability of the event in each interval and the cumulative survival.
  • Likelihood ratio test (LRT)conceptA test that compares the maximized likelihoods of two nested models to decide whether the added parameters improve the model significantly.
  • Likert Scale AnalysisanalysisAn analysis that summarises, item by item, how responses are distributed on Likert-type items running from 'strongly disagree' to 'strongly agree', the shares agreeing and disagreeing, and the consistency of the item set.
  • Limits of agreement (Bland-Altman)conceptThe range within which about 95% of the differences between two measurement methods applied to the same units are expected to fall.
  • Lin's concordance correlation coefficient (CCC)conceptAn agreement coefficient that measures not only whether the values from two measurement methods vary together but also how close they fall to the 45-degree line of equality.
  • Linear Mixed Models (LMM)analysisA regression method for dependent data, such as repeated measurements of the same person or observations within the same school, hospital or plot, that models a continuous outcome with fixed effects common to the whole sample and random effects that vary from group to group.
  • Link functionconceptIn generalized linear models, the function that connects the expected value of the response to a linear combination of the predictors: g(μ) = Xβ.
  • Linkage method (Ward, complete, average, single)conceptIn hierarchical clustering, the rule that determines how the distance between two clusters is computed from the distances between their members.
  • Listwise and pairwise deletionconceptTwo simple ways of handling observations with missing data: dropping any row with a missing value on any variable (listwise), or, for each calculation, dropping only rows missing on the variables that calculation uses (pairwise).
  • Ljung-Box (portmanteau) testconceptA test of whether the autocorrelations of a time series, or of a model's residuals, over the first h lags are jointly zero.
  • Lo-Mendell-Rubin (LMR) testconceptAn adjusted likelihood-ratio test in latent class and latent profile analysis of whether a model with K classes fits the data significantly better than one with K − 1 classes.
  • Local indicators of spatial association (LISA) and Gi*conceptLocal statistics that break global spatial autocorrelation down to individual locations, showing whether each unit forms a significant cluster of high or low values with its surroundings.
  • Local outlier factor (LOF)conceptA method that finds outliers even in regions of the data with different densities, by comparing the density around an observation with the density around its neighbours.
  • LOESS (Local Regression)analysisA method that shows how an outcome changes with one or two explanatory variables through a smooth curve (or surface) built from small regressions fitted locally to the data, without assuming in advance a shape such as a line or a parabola.
  • Log returnconceptThe natural logarithm of the ratio of two successive values in a price series: r_t = ln(P_t / P_(t-1)); it is usually multiplied by 100 to put it on a percentage-like scale.
  • Log-Linear ModelanalysisA method that models the cell counts of a crosstab formed by three or more categorical variables to find out which variables are associated and whether those associations change across the levels of the others.
  • Log-rank test (and Breslow, Tarone-Ware)conceptA test of whether the survival curves of two or more groups are equal, comparing observed and expected numbers of events at each event time.
  • Logistic RegressionanalysisA regression model that predicts the probability of a two-category (yes/no) outcome from one or more predictors and expresses the effect of each predictor as an odds ratio (OR).
M
  • M-estimator (Huber, Tukey biweight)conceptA family of estimators of location or regression that resist outliers by using a loss function that limits the influence of large residuals.
  • Mahalanobis distanceconceptA distance that measures how far an observation lies from the centre of a multivariate distribution, taking into account the scales of the variables and the correlations between them.
  • Mann-Kendall + Sen's SlopeanalysisA non-parametric method that tests whether a time series shows a steady increase or decrease over time (a monotonic trend) without distributional assumptions, and gives the size of the trend as Sen's slope.
  • Mann-Whitney UanalysisA test that needs no normality: it converts the values of two independent groups to ranks and compares them, testing whether one group's values tend systematically to be larger than the other's.
  • MANOVAanalysisA multivariate analysis of variance that tests in a single test whether groups differ on several continuous outcomes taken together, that is, on the profile formed by the means of those outcomes.
  • Mantel-Haenszel common odds ratioconceptAn estimator that combines 2x2 tables, split by the strata of a confounding variable, into a single odds ratio assumed to be common to all strata.
  • Map classification methods (equal interval, quantile, natural breaks)conceptRules that decide how continuous values on a thematic map are divided into colour classes.
  • Mardia's test of multivariate normalityconceptA test of whether several variables jointly follow a multivariate normal distribution, based on measures of multivariate skewness and kurtosis.
  • Marginal and conditional maximum likelihood (MML, CML)conceptTwo methods in item response theory that estimate item parameters either by integrating person abilities out over a distribution (MML) or by conditioning on a sufficient statistic (CML).
  • Mauchly's test of sphericityconceptA test, used in repeated measures ANOVA, of the assumption that the variances of all pairwise differences between conditions are equal (sphericity).
  • Maximum likelihood (ML) estimationconceptAn estimation method that chooses the model parameters as the values that make the observed data most probable (maximize the likelihood).
  • Maximum tree depthconceptThe maximum number of levels of splits allowed from the root to the furthest leaf of a decision tree; a hyperparameter that limits the tree's complexity.
  • McDonald's ω (omega) reliability coefficientconceptAn internal-consistency reliability coefficient that estimates, from factor loadings, how much of the variance in a scale's total score is due to the common factor(s).
  • McNemar TestanalysisA test of whether the proportion of a yes/no outcome measured twice on the same people changes between the two measurements, based only on those whose answer or status changed.
  • Measurement invariance (configural, metric, scalar, strict)conceptTesting, with progressively stricter equality constraints, whether an instrument measures the same construct in the same way across groups (sex, country, time).
  • Measurement model and structural modelconceptIn structural equation modelling, the part that defines how latent variables are measured by observed indicators (the measurement model) and the part that defines the directed relationships among latent variables (the structural model).
  • Median absolute deviation (MAD) and modified z-scoreconceptAn outlier-resistant measure of spread equal to the median of the absolute distances of the observations from their median; the modified z-score is the standardized distance computed with it.
  • Median survival time and restricted mean survival time (RMST)conceptTwo summary measures in survival analysis: the median is the time by which half of the participants have had the event, and the RMST is the average event-free time up to a chosen time limit.
  • Mediation AnalysisanalysisAn analysis that shows how much of a variable's (X) effect on an outcome (Y) is transmitted through a third variable (the mediator, M), by estimating the indirect effect and its confidence interval.
  • Meta-AnalysisanalysisA method that combines the results of several studies on the same question into a single overall effect estimate, weighting each study by its precision, and examines how the studies differ.
  • Meta-regressionconceptA regression model used in meta-analysis to explain differences in effect sizes across studies by study-level characteristics (moderators).
  • Metric and non-metric MDS (SMACOF)conceptThe two forms of multidimensional scaling, which displays distances or dissimilarities between objects on a low-dimensional map, preserving either the distances themselves (metric) or only their rank order (non-metric).
  • Mid-p valueconceptA less conservative p-value for exact tests, obtained by counting only half of the probability of the observed result.
  • Minimum covariance determinant (MCD) and robust distanceconceptA robust estimator for multivariate data that uses the 'tightest' subset of observations to obtain location and covariance estimates unaffected by outliers; the Mahalanobis distance computed from it is called the robust distance.
  • Minimum leaf / node sizeconceptIn decision trees, the minimum number of observations a leaf must contain (or a node must contain to be split); a setting that limits how finely the tree can grow.
  • Missing-data mechanisms (MCAR, MAR, MNAR)conceptRubin's (1976) three-way classification of how the probability that a value is missing relates to the observed and unobserved values in the data.
  • Mixed-Design ANOVA, Split-PlotanalysisAn analysis of variance for designs in which different groups are each measured several times, testing together the group difference, the change over time and, above all, whether the groups change differently over time (the group × time interaction).
  • Model fit indices (χ², CFI, TLI, RMSEA, SRMR)conceptMeasures that summarize, from different angles, how well the covariance structure implied by a confirmatory factor or structural equation model matches the observed data.
  • Model identification (std.lv, unit loading)conceptThe condition that a model's parameters can be estimated from the observed data as a single set of values; in latent variable models this requires constraints such as fixing the scale.
  • Moderator variableconceptA third variable that changes the direction or strength of the relationship between two variables; its effect is usually modelled with an interaction term.
  • Modification indicesconceptIn structural equation models, statistics showing roughly how much the model chi-square would drop if a parameter currently fixed at zero were freed.
  • Monte Carlo (simulation) envelopeconceptA band built from the values of a summary function (e.g. Ripley's K or L) in many simulations under the null hypothesis, used to check visually whether the observed curve is consistent with chance.
  • Monte Carlo p-valueconceptAn approximate p value obtained by mimicking the distribution of a test statistic under the null hypothesis through a large number of random rearrangements or simulations.
  • Mood's median testconceptA nonparametric test of whether two or more independent groups share the same median, based on whether observations fall above or below the overall median.
  • Moran's I (Spatial Autocorrelation)analysisA method that tests whether a variable takes similar values at nearby locations, that is, whether there is spatial autocorrelation, both with a single statistic for the whole area (global Moran's I) and separately for each location (local Moran, LISA).
  • Moving-average and exponential moving-average smoothingconceptSmoothing methods that suppress short-term fluctuations in a time series by averaging nearby observations at each point, making the trend visible.
  • MR Frequency AnalysisanalysisAn analysis summarising, for a multiple-response question where a person may tick more than one option, how many times each option was chosen and what percentage of respondents chose it.
  • MR × Categorical CrosstabanalysisAn analysis of whether the rate at which each option of a multiple-response question is chosen differs between the groups of a single-response categorical variable (such as sex, age group or education level).
  • MR × MR CrosstabanalysisAn analysis that examines, cell by cell, the association between the options of two separate multiple-response questions by looking at how often each pair of options is chosen together.
  • mtry (variables tried per split)conceptIn a random forest, the number of variables chosen at random as candidates at each split point of the trees.
  • Multicollinearity (VIF, tolerance, condition index)conceptStrong linear relationships among the explanatory variables in a regression, which make their coefficients unstable and imprecisely estimated.
  • Multidimensional Scaling (MDS)analysisA method that starts from information on the distances or similarities between objects and draws a 'similarity map' in two or a few dimensions that distorts those distances as little as possible.
  • Multinomial Logistic RegressionanalysisA regression model that relates which of three or more unordered categories an outcome falls into (e.g. mode of transport, programme choice) to predictors, comparing each category with a reference category.
  • Multiple Comparison CorrectionanalysisMethods that adjust p values or the decision threshold when many hypothesis tests are run in the same study, to stop results that are 'significant' by chance from piling up.
  • Multiple comparisons and the familywise error rate (FWER)conceptThe probability of wrongly rejecting at least one true null hypothesis within a family of tests; multiple-comparison corrections exist to control it.
  • Multiple ImputationanalysisA method that fills in missing values many times, each time with slightly different predictions based on the observed data, repeats the analysis in every completed data set and combines the results so that they also reflect the uncertainty created by the missingness.
  • Multiple imputation by chained equations (FCS/MICE) and PMMconceptA multiple imputation approach that fills in missing values sequentially and iteratively with a separate model predicting each variable from the others, producing several completed data sets.
  • Multiple Linear RegressionanalysisA method that models how a continuous outcome relates to several predictors, estimating the average change in the outcome associated with each predictor while the others are held constant.
  • Multiple marginal independence tests (MMI, SPMI)conceptMethods that use a single overall test to check whether a multiple-response question (where several options can be ticked) is associated with another variable or with another multiple-response question.
  • Multiple-response set (MR set)conceptA set of columns for a question where respondents can tick several options, with each option stored in its own column and the set treated as one variable in the analysis.
  • Multivariate test statistics (Wilks' Λ, Pillai's trace, Hotelling-Lawley, Roy)conceptFour classical test statistics that compare the mean vectors of several dependent variables across groups at once.
N
  • Nested and crossed random effectsconceptTwo structures describing how two grouping factors relate in mixed models: in a nested structure each level of one factor occurs within only one level of the other, while in a crossed structure their levels can combine in any way.
  • Nested Mixed Model (Nested LMM, Multi-Level Hierarchical Mixed Model)analysisA mixed model for hierarchical trials in which subgroups sit within only one higher group (farms within a region, blocks within a site), splitting the variability in the outcome by level and testing each effect against the correct error term.
  • Neural Network (MLP)analysisA flexible model that predicts a class or a numeric value by passing the inputs through intermediate layers of weighted sums and non-linear transformations, learning its weights from the data so as to minimise the error.
  • Node impurity (Gini, entropy)conceptA criterion that measures how mixed the classes are among the observations in a decision-tree node and is used to choose the best split.
  • Non-informative (flat, Jeffreys) priorconceptA prior distribution in Bayesian analysis that carries no strong prior belief about a parameter and aims to let the results be determined mainly by the data.
  • Noncentral t confidence intervalconceptAn exact confidence interval for standardized effect sizes such as Cohen's d, built by searching for the noncentral t distributions that could have produced the observed t value.
  • Nonlinear least squares and starting valuesconceptA method that fits a model that is nonlinear in its parameters by iteratively minimizing the sum of squared differences between observed and predicted values; the solution is searched for starting from given initial values.
  • Nonlinear RegressionanalysisA regression that models the relation between an outcome and a predictor with a function whose parameters enter nonlinearly, often one derived from theory (exponential, logistic, Michaelis-Menten and so on), and estimates those parameters by least squares.
  • Normality assumptionconceptThe assumption, made by many parametric tests, that the data or the model residuals come from a normal (Gaussian) distribution.
  • Normality TestsanalysisTests and plots that check whether the values of a numeric variable could plausibly have come from a normal (bell-shaped) distribution.
  • Nugget, sill and rangeconceptThe three parameters that define a variogram model: the unexplained variability at very short distances (nugget), the plateau the variogram reaches (sill), and the distance at which spatial correlation dies out (range).
  • Null hypothesis (H₀) and test valueconceptThe statement of 'no difference', 'no effect' or 'the parameter equals a given value' that a hypothesis test assumes, and against which the data are checked for incompatibility.
  • Number needed to treat (NNT)conceptThe number of people who must receive a treatment, compared with the control condition, to produce one additional good outcome (or prevent one additional bad event).
  • Number of nearest neighbours (k) and the neighbourhood graphconceptThe number of closest observations considered around a given observation, and the network that links each observation to those neighbours; together they define local structure in classification, outlier detection and dimension reduction.
  • Number of trees (ensemble models)conceptThe number of decision trees grown in ensemble methods such as random forests, gradient boosting and Isolation Forest; it plays a different role depending on the method.
O
  • Observation window (study region)conceptIn point pattern and density analyses, the boundary of the region in which points could be observed and in which the analysis is valid.
  • Observed (post hoc) powerconceptThe statistical power of a test computed by treating the effect size observed in the sample as if it were the true effect.
  • Odds ratio (OR)conceptThe ratio of the odds of an event in one group (the probability it happens divided by the probability it does not) to its odds in another group.
  • Offset (exposure term)conceptA term added to the linear predictor of a count model with its coefficient fixed at 1 (usually the logarithm of exposure), to account for observations being counted over different lengths of time, areas or populations.
  • Omega squared (ω²)conceptAn ANOVA effect size that estimates the proportion of variance explained by an effect in the population, with less bias than eta squared.
  • One-Sample t-TestanalysisA test of whether the mean of a sample truly differs from a pre-specified value (a target, a norm, or the value from an earlier study).
  • One-sided and two-sided tests (hypothesis direction)conceptThe choice of whether the alternative hypothesis looks for a difference in one direction only (one-sided) or in either direction (two-sided).
  • One-standard-error (1-SE) ruleconceptIn cross-validation, the rule of choosing the simplest model among those whose error lies within one standard error of the error of the best model.
  • One-Way ANOVAanalysisAn analysis of variance that tests whether the means of three or more independent groups are equal, by comparing the differences between the groups with the spread within them.
  • Optimization algorithm (Adam, SGD, L-BFGS)conceptA numerical method that updates a model's parameters step by step so as to minimize the loss function (or the residual sum of squares).
  • Ordinal association measures (gamma, tau-b, tau-c, Somers' d)conceptCoefficients that measure the direction and strength of the association between two ordinal categorical variables from the counts of concordant and discordant pairs of observations.
  • Ordinal Logistic RegressionanalysisA regression model that relates the tendency of an ordered categorical outcome, such as low-medium-high or Likert responses, to fall in higher categories to predictors, giving the effects as odds ratios.
  • Ordinary least squares (OLS)conceptA method that estimates regression coefficients by minimizing the sum of squared differences between the observed values and the model's predictions (the residuals).
  • OutlierconceptAn observation that lies clearly apart from the rest of the data and may reflect a measurement error, a different process or a genuine but rare case.
  • Overdispersion and the scale parameterconceptThe situation in which the observed variance of count or proportion data exceeds what the Poisson or binomial model predicts, and the scale (dispersion) parameter used to correct for it.
P
  • p-valueconceptThe probability, assuming the null hypothesis is true, of obtaining a result at least as extreme as the one observed purely by chance.
  • PABAK and κ_maxconceptTwo companion measures that help interpret a Cohen's kappa that is misleadingly low because categories are very unevenly distributed or the raters have different tendencies.
  • Page's test (ordered alternatives)conceptA nonparametric test of whether three or more conditions measured on the same subjects show an increasing (or decreasing) trend in a pre-specified order.
  • Paired t-TestanalysisA test of whether the mean difference between two measurements taken on the same people, or on matched units, truly differs from zero.
  • Panel Data RegressionanalysisRegression methods for data in which the same units (firms, households, provinces, countries) are followed over several periods, estimating relationships between variables while controlling for unobserved, time-invariant characteristics of the units.
  • Parallel analysis (Horn) and the MAP criterionconceptA method for deciding the number of factors or components by comparing the observed eigenvalues with eigenvalues from random data, and a further criterion based on partial correlations.
  • Parallel and strictly parallel measures modelconceptA classical test theory model that assumes all items in a scale measure the same true score with the same sensitivity and the same amount of error (and, in the strictly parallel model, also have equal means).
  • Parallel trends assumptionconceptThe key assumption of difference-in-differences: without the intervention, the average outcomes of the treated and untreated groups would have changed by the same amount over time.
  • Parametric Survival (AFT)analysisA survival model that describes the time until an event with a specific probability distribution, such as Weibull or log-normal, and gives by what factor each factor lengthens or shortens that time (the time ratio).
  • Partial and semi-partial correlationconceptCorrelations that show how strong the relationship between two variables is after the influence of one or more third variables has been removed.
  • Partial AUCconceptThe area under the ROC curve computed over a restricted range of specificity (or sensitivity) only, for example where specificity exceeds 0.90.
  • Partial dependence plotconceptA plot showing how a machine-learning model's average prediction changes as one predictor is varied, averaging over the effects of the other predictors.
  • Partial Least Squares Regression (PLS)analysisA regression method that reduces many strongly correlated predictors to a few latent components most related to the response, and models the response with those components.
  • Path AnalysisanalysisAn analysis that estimates a theoretical model of directed relationships among observed variables (who affects whom) as a system of simultaneous equations and tests how well the model fits the data.
  • Pearson correlation coefficientconceptA coefficient between -1 and +1 that measures the direction and strength of the linear relationship between two numerical variables.
  • Percent of cases and percent of responsesconceptTwo kinds of percentage for multiple-response questions, dividing an option's count either by the number of respondents (percent of cases) or by the total number of selections (percent of responses).
  • Percentiles (quantiles) and percentile definitionsconceptThe value below which a given share of the ordered data falls; the 90th percentile, for example, is the point below which about 90% of the observations lie.
  • Periodogram and spectral densityconceptA tool that shows how the total variance of a time series is distributed across frequencies (cycle lengths), revealing the dominant periods in the series.
  • Permutation importanceconceptA model-agnostic importance measure that gauges a variable's contribution to prediction by how much the model's test performance drops when that variable's values are randomly shuffled.
  • Permutation p-valueconceptA p value based on the distribution of a test statistic when group labels are shuffled at random, requiring no distributional assumption.
  • Permutation TestanalysisA test that needs no distributional assumptions: it builds from the data itself the distribution of 'what would we see if there were really no difference?' by randomly reassigning group labels (or the signs of paired differences), and computes the p value from that distribution.
  • Perplexity (t-SNE effective number of neighbours)conceptThe t-SNE parameter that sets the width of each point's neighbourhood and roughly indicates how many effective neighbours each point takes into account.
  • Person ability (θ) estimation (EAP, MAP, ML, WLE)conceptMethods in item response theory that estimate each person's latent ability or trait level (θ) from their response pattern, given the item parameters.
  • Person-time (person-years) denominatorconceptThe sum of the time each individual in a study spends at risk; used as the denominator of event rates (incidence rate, mortality rate).
  • Phi (φ) coefficientconceptA measure of association that expresses the strength of the relationship between two binary (two-category) variables on a scale from -1 to +1.
  • PLS-SEM (Partial Least Squares Path Modeling)analysisA variance-based structural equation method that estimates path relations between indicator-measured latent constructs by iteratively building construct scores and running regressions on them, aiming to maximise the explained variance of the endogenous constructs.
  • Point Pattern Analysis (Ripley K/L, G/F/J, CSR)analysisMethods (Ripley's K/L, G/F/J, Clark-Evans, the quadrat test) that test whether point locations within an area are randomly scattered or more clustered or more regular than expected, and at which distance scale this holds.
  • Poisson / Negative Binomial Regression (Count Regression)analysisRegression models that show how count outcomes such as 0, 1, 2, … accidents, visits, species or errors change with predictors, expressing each effect as a rate ratio (IRR).
  • Polychoric and tetrachoric correlationconceptAn estimate of the correlation between the continuous, normally distributed latent variables assumed to underlie two ordinal (or binary) variables.
  • Polynomial (trend) contrastsconceptMutually orthogonal contrasts that test whether the means follow a linear, curved (quadratic) or more complex trend across the levels of an ordered factor.
  • Pooled variance (pooled standard deviation)conceptA single common variance estimate obtained, under the assumption of equal group variances, by combining the group variances weighted by their degrees of freedom.
  • Positive and negative predictive values (PPV, NPV)conceptThe probability that someone is truly positive when a diagnostic test or classifier says 'positive' (PPV), and truly negative when it says 'negative' (NPV).
  • Post hoc tests (pairwise comparisons)conceptComparisons that, once an omnibus test (such as ANOVA) has found a difference among groups, identify which pairs differ while controlling the multiple-testing error.
  • Posterior distribution and posterior predictive checkconceptIn Bayesian analysis, the probability distribution obtained by combining prior information with the data, which expresses updated belief about the possible values of a parameter; the posterior predictive check tests the model's fit with data generated from this distribution.
  • Precision-recall curve and average precisionconceptA curve showing a classifier's precision (how many positive predictions are correct) against its recall (how many true positives are caught) across thresholds, and average precision (AP), which summarises it in a single number.
  • Prediction error measures (RMSE, MAE, MAPE, MASE)conceptMeasures that summarize how far a model's predictions deviate from the actual values.
  • Prediction intervalconceptThe range expected, with a given probability, to contain a single new observation (or the true effect of a new study), rather than the mean response.
  • Primary sampling unit (PSU) and cluster samplingconceptIn multistage sampling, the units selected at the first stage (e.g. neighbourhoods, schools); because observations from the same unit resemble each other, variance is computed over these units.
  • Principal Component Analysis (PCA)analysisA dimension reduction method that condenses many correlated numeric variables into a few uncorrelated components, ordered by importance, while keeping most of the information.
  • Prior class probabilities (discriminant analysis)conceptIn classification, the assumed probability that an observation belongs to each class before the data are seen; these probabilities also shape the classification decision and its threshold.
  • Prior distributionconceptIn Bayesian analysis, the probability distribution expressing the information or uncertainty about a parameter before the data are seen.
  • Probit RegressionanalysisA regression model that links the probability of a binary outcome (died / survived, passed / failed) to predictors through the cumulative standard normal distribution, widely used in dose-response studies to estimate effective doses (ED50, LD50).
  • Profile likelihood confidence intervalconceptA method that derives a parameter's confidence interval from the profile likelihood curve, in which the other parameters are re-estimated to maximize the likelihood at each value of that parameter.
  • Propensity Score AnalysisanalysisA method that estimates the average effect of a treatment in observational data without random assignment, by making treated and untreated individuals similar in their observed characteristics (matching, weighting, subclassification).
  • Propensity score matching (nearest neighbour, caliper)conceptA method for observational data that builds comparable groups by matching treated and control units whose probability of receiving treatment (the propensity score) is similar.
  • Propensity score subclassification (stratification)conceptA method that divides observations into a few strata by their propensity scores, estimates the treatment effect within each stratum and then combines them.
  • Proportion of explained varianceconceptThe share of the total variability in the data captured by a component, factor or model.
  • Proportional biasconceptA situation in which the difference between two measurement methods is not constant but grows or shrinks with the size of the quantity being measured.
  • Proportional hazards assumptionconceptThe assumption in the Cox model that the risk ratio between two subjects or groups stays constant over time.
  • Proportional odds (parallel lines) assumptionconceptIn ordinal logistic regression, the assumption that each predictor's effect (slope) is the same at every threshold of the outcome variable.
  • Proportional reduction in error measures (lambda, Goodman-Kruskal τ, uncertainty coefficient)conceptMeasures of association that show by what percentage knowing one nominal variable reduces the error in predicting a second nominal variable.
  • Pseudo-R² (McFadden, Cox-Snell, Nagelkerke)conceptMeasures that, by analogy with R² in linear regression, summarize how much a model estimated by maximum likelihood, such as logistic regression, improves on the null model.
  • Publication biasconceptThe tendency of the published literature to overstate the true effect because studies with significant or expected results are more likely to be published.
Q
  • Q-Q and P-P plotsconceptPlots that show how well data follow a theoretical distribution (usually the normal) by comparing observed and expected quantiles (Q-Q) or cumulative probabilities (P-P).
  • QIC and QICu criteriaconceptQuasi-likelihood-based model selection criteria used in generalized estimating equations (GEE) in place of the AIC, since no full likelihood is defined there.
  • Quadrat count testconceptA chi-square test of whether points are distributed according to complete spatial randomness (CSR), carried out by dividing the study area into equal cells (quadrats) and comparing the number of points in each.
  • Quantile RegressionanalysisA regression that estimates how chosen quantiles of the outcome, such as the median, the lower quartile or upper percentiles, rather than its mean, change with the predictors, showing whether an effect differs across parts of the distribution.
  • Q² predictive relevance (Stone-Geisser, PRESS)conceptAn R²-like measure based on cross-validation that shows how well a model predicts data points not used in its estimation.
R
  • r effect size from Z/√NconceptAn effect size for rank-based tests, obtained by dividing the test's standard normal statistic Z by the square root of the sample size, and read on the correlation scale.
  • Ramsey RESET testconceptA general specification test of whether the functional form of a linear regression model is wrong, carried out by adding powers of the fitted values to the model.
  • Random ForestanalysisA machine learning method that predicts a class or a numeric value by growing many decision trees on different resamples of the data and random subsets of the variables, then combining the trees' votes (or averaging their predictions).
  • Random seedconceptThe number that sets the starting point of the computer's random number generator, so that the same 'random' results are obtained again whenever the same value is used.
  • Random slopeconceptA random effect in mixed models that lets the effect (slope) of a predictor on the outcome vary from group to group.
  • Random starts and local optimaconceptRunning an iterative algorithm many times from different starting points, so that it does not get stuck in a merely local optimum and has a better chance of reaching the true best solution.
  • Rank-biserial correlationconceptA rank-based effect size ranging from -1 to +1 that accompanies the Mann-Whitney U or Wilcoxon signed-rank test.
  • Rate ratio and rate differenceconceptTwo measures comparing the event rates (events per unit of person-time) of two groups: the ratio of the rates (rate ratio) and their difference (rate difference).
  • Ratio estimator (Y/X)conceptAn estimator for survey data that estimates the ratio of two totals, such as expenditure per person or cars per household, as the ratio of weighted totals.
  • Redundancy indexconceptA measure in canonical correlation analysis of how much of the total variance in one set of variables is explained by a canonical variate of the other set.
  • Reference category (comparison level)conceptThe level against which the other levels are compared when a categorical variable enters a model as dummy variables, and whose own coefficient is fixed at zero.
  • Regression coefficient (B) and parameter estimatesconceptThe estimated average change in the outcome produced by a one-unit increase in a predictor, holding the other predictors constant.
  • Regression through the origin (no intercept)conceptA model that drops the intercept and forces the regression line or plane through the origin (zero response when all predictors are zero).
  • Regularization (penalty) and the penalty parameter λ/αconceptA method that reduces overfitting and stabilizes estimates by adding a penalty on the size of the model coefficients; the strength of the penalty is set by λ (α in some software).
  • Regularized Regression (LASSO / Ridge / ElasticNet)analysisRegression methods (LASSO, Ridge, ElasticNet) that prevent overfitting when there are many correlated predictors by penalising the size of the coefficients, and that, in variants such as LASSO, also select predictors.
  • Relative risk (RR) and risk difference (RD)conceptThe ratio (RR) and the difference (RD) of the probabilities of an event in two groups.
  • Repeated Measures ANOVAanalysisAn analysis of variance for designs in which the same people are measured at three or more times or under three or more conditions, testing whether the means differ after setting aside the lasting differences between people.
  • Residual plots and residual diagnosticsconceptChecking a model's assumptions and fit by examining its residuals (observed minus predicted values) with plots and summary statistics.
  • Response Surface Regression (RSREG)analysisA method that models how an outcome in a designed experiment changes with two or more numeric factors using a second-order surface, and searches for the factor levels that make the outcome highest or lowest.
  • Restricted maximum likelihood (REML)conceptA form of maximum likelihood that estimates variance components in mixed models while accounting for the degrees of freedom used up by estimating the fixed effects.
  • Reverse-coded itemsconceptRecoding the scores of items worded in the opposite direction to the rest of a scale so that all items point the same way.
  • Ridge analysis (response surface)conceptA method for a second-order response surface model that traces the factor settings giving the largest (or smallest) predicted response on spheres of increasing radius around the centre of the experimental region.
  • Ridge RegressionanalysisA regression method that stabilises the coefficients when predictors are strongly correlated by shrinking them towards zero with a small penalty, with the aim of better prediction on new data.
  • Ripley's K/L and the G, F, J functionsconceptSummary functions that examine whether a point pattern is clustered, random or regular at different distance scales by comparing it with complete spatial randomness (CSR).
  • Robust (sandwich) standard errors (HC0-HC3)conceptCoefficient standard errors that remain consistent even when the error variance is not constant or the model's variance structure is misspecified.
  • Robust ML estimators (MLM, MLR; Satorra-Bentler)conceptEstimators for structural equation and confirmatory factor models that keep the ML estimates but correct the standard errors and the chi-square test when the data are not normally distributed.
  • Robust Regression (M-estimator)analysisA family of linear regression methods that stops outliers, and leverage points with unusual predictor values, from distorting the coefficients by limiting their influence.
  • Robust regression estimators (MM, S, LTS) and the breakdown pointconceptEstimators that limit how far outlying observations and leverage points can distort regression coefficients, and the breakdown point, which shows how much contamination an estimator can withstand.
  • ROC CurveanalysisA curve showing how well a continuous test value or model probability distinguishes two states (ill/healthy, event/no event) across all possible thresholds, and the area under the curve (AUC) that summarises this discrimination in a single number.
  • Rubin's rules (pooling imputation results)conceptRules that combine the results of analyses run separately on each of the m completed data sets from multiple imputation into a single estimate with the correct uncertainty.
  • Runs TestanalysisA nonparametric test of whether the observations in a sequence arrive in random order, or whether similar values cluster together or alternate in a regular way.
S
  • Sample Selection and Truncated Regression (Heckman)analysisMethods that estimate regression coefficients while correcting the bias arising when the outcome is observed only for a selected subgroup (the Heckman model) or when observations beyond a certain limit never enter the data at all (truncated regression).
  • Sample selection bias (Heckman)conceptBias in estimates that arises because the outcome is observed only for units that pass through a selection process related to the outcome; the Heckman model was developed to correct it.
  • Sampling weight (design weight)conceptIn complex samples, a weight showing how many population units each observation represents, usually defined as the inverse of its probability of selection.
  • SAR (Spatial Autoregressive Lag)analysisA spatial regression model for situations where the outcome in one area is also affected by the outcomes in neighbouring areas, estimating this spillover (spatial lag) together with the effects of the explanatory variables.
  • Satterthwaite and Kenward-Roger degrees-of-freedom approximationsconceptMethods that compute approximate (often non-integer) degrees of freedom for t and F tests when the variance of an estimate is a combination of several variance components.
  • Scatter plotconceptA graph that shows each observation of two numerical variables as a point whose coordinates are its two values.
  • Scheffé testconceptA multiple-comparison method used after ANOVA that tests not only pairwise comparisons but all possible linear contrasts of the group means at once, while controlling the familywise error rate.
  • Score (Lagrange multiplier) testconceptA test of whether a parameter restriction (e.g. coefficient = 0) holds that fits only the restricted model and uses the slope of the likelihood at that point.
  • Score and scale tests (Van der Waerden, Savage, Siegel-Tukey, Ansari-Bradley, Mood, Klotz)conceptNonparametric tests that use scores derived from ranks, instead of raw ranks, to test for differences in location or spread between groups.
  • Scree plotconceptA plot of the eigenvalues from a principal component or factor analysis in decreasing order, used to help decide how many components or factors to keep.
  • Seasonal Kendall testconceptThe seasonal extension of the Mann-Kendall test, which tests for a monotonic trend in seasonal time series by comparing each season only with itself.
  • Seasonality and seasonal periodconceptA pattern in a time series that repeats regularly with a calendar cycle of fixed length (day, week, year); the length of the cycle in number of observations is the seasonal period (m).
  • Seemingly unrelated regression (SUR) and 3SLSconceptSystem methods that estimate several regression equations, each with its own outcome, jointly by exploiting the correlation between the equations' errors.
  • Selection stabilityconceptA method for judging how reliable a variable selection is by measuring how often a selecting method such as the lasso picks the same variables across different subsamples of the data.
  • Sen's slopeconceptA robust method that estimates the size of a linear trend in a time series as the median of the slopes between all pairs of observations.
  • Sensitivity analysisconceptAn approach that checks how robust the result of an analysis is to reasonable changes in the data, the assumptions or the methodological choices.
  • Sensitivity and specificityconceptThe proportion of true positives that a diagnostic test or classifier correctly detects (sensitivity) and the proportion of true negatives it correctly rules out (specificity).
  • Serial correlation tests (Breusch-Godfrey)conceptTests of whether successive error terms over time or sequence are correlated with one another; the Breusch-Godfrey test is one of the most general of these.
  • SHAP valuesconceptA measure, based on Shapley values from game theory, that explains each individual prediction of a machine-learning model as the sum of the contributions of the variables.
  • Shapiro-Wilk testconceptA test of whether a sample comes from a normal distribution, based on how well the sorted data agree with the ordered values expected under normality.
  • Sign TestanalysisA test that looks only at the direction (plus or minus) of the differences between two measurements on the same people, checking whether increases and decreases are equally frequent, that is, whether the median difference is zero.
  • Significance level (α)conceptThe upper limit, set by the researcher in advance, on the probability of rejecting a null hypothesis that is actually true.
  • Silhouette coefficientconceptAn index between -1 and +1 that measures how well each observation fits its own cluster compared with the nearest neighbouring cluster, summarizing clustering quality.
  • Simple effectsconceptTesting the effect of one factor separately at each level of the other factor; the basic way to unpack an interaction.
  • Single imputation (mean, median, kNN filling)conceptCompleting the data by replacing each missing value with a single estimated value (the column mean or median, or the average of similar observations).
  • SkewnessconceptA measure of whether a distribution is symmetric around its mean and in which direction its tail stretches.
  • Smoothing parameter (λ, span)conceptThe setting in curve-fitting methods that controls how flexible or smooth the curve is, that is, the balance between closeness to the data and smoothness.
  • Sobel testconceptA method in mediation analysis that tests whether the indirect effect (a × b) differs from zero with a z test based on the approximate standard error of the product.
  • Spatial autocorrelation (Moran's I, Geary's C)conceptThe tendency of a variable to take similar (or dissimilar) values at nearby locations; its common measures are Moran's I and Geary's C.
  • Spatial Error Model (SEM)analysisA spatial regression model for when the similarity of neighbouring areas comes from shared factors left out of the model, which models this dependence in the error term and estimates the effects of the explanatory variables with correct standard errors.
  • Spatial lag and spatial Durbin models (SAR, SDM, SDEM)conceptRegression models that use a spatial weights matrix W to model how the outcome at one location is influenced by outcomes, or by explanatory variables, at neighbouring locations.
  • Spatial weights matrix (W)conceptThe n × n matrix that defines, in spatial analysis, which units count as neighbours and how much weight each neighbour relationship carries.
  • Spearman rank correlationconceptA correlation coefficient that measures the strength and direction of a monotonic relationship between two variables using the ranks of the values rather than the values themselves.
  • Spearman-Brown formula (single and average measures)conceptA formula that predicts how the reliability of a measurement changes when the number of comparable measurements is increased or reduced: ρ_k = k·ρ / (1 + (k - 1)·ρ).
  • Sphericity assumptionconceptThe assumption in repeated measures ANOVA that the variances of the difference scores between all pairs of levels of a within-subject factor are equal.
  • Splines and basis dimension (k)conceptFunctions that model a variable's effect flexibly with piecewise polynomials joined smoothly at knots, and the number of basis functions that caps this flexibility.
  • Split-half reliability and Guttman's λconceptA method that estimates the reliability of a whole scale from the relationship between the total scores of its two halves; Guttman's λ coefficients are generalized lower bounds built on the same idea.
  • Standard errorconceptThe standard deviation of an estimate's sampling distribution, showing how much an estimate (such as a mean or a regression coefficient) would vary from sample to sample.
  • Standard error of measurement (SEM) and minimal detectable change (MDC)conceptA value showing how much a person's observed score typically varies around their true score because of measurement error, and the smallest difference that can be counted as real change derived from it.
  • Standardized and adjusted residuals (contingency tables)conceptValues showing, on a comparable scale, how far the observed count in each cell of a contingency table departs from the expected count.
  • Standardized mean difference (SMD) and covariate balanceconceptThe difference between two group means expressed in standard deviation units; in propensity score analysis, a measure of how well balanced the groups are on a variable.
  • Standardized regression coefficient (β)conceptThe coefficient obtained when all variables are expressed in standard deviation units; it shows how many standard deviations the outcome changes for a one-standard-deviation increase in the predictor.
  • State-space model and the Kalman filterconceptA modelling framework that describes a time series as an unobserved, time-varying state vector (level, slope, season) plus a noisy measurement of it, together with the algorithm that estimates the state step by step.
  • StationarityconceptThe property of a time series whose statistical characteristics, its mean, variance and lag-to-lag dependence, do not change over time.
  • Stationary point and canonical analysis (response surface)conceptThe point on a second-order response surface where all slopes are zero, and the analysis that uses eigenvalues to determine whether that point is a maximum, a minimum or a saddle point.
  • Stepwise variable selectionconceptAn automatic variable-selection method that adds candidate variables to a model, or removes them, one step at a time according to a statistical criterion.
  • STL Seasonal DecompositionanalysisA method that splits a time series into its long-term direction (trend), a regularly repeating seasonal pattern and the remaining irregular fluctuation (remainder).
  • Stratified analysis (stratum variable)conceptAn approach that controls for a confounding or design variable by running the analysis separately within each of its levels (strata) and then combining the results.
  • Stratified sampling (design stratum)conceptA sampling design in which the population is first divided into mutually exclusive subgroups (strata) and a sample is drawn independently from each stratum.
  • Stress, fit measures and the Shepard diagram (MDS)conceptMeasures, and the accompanying plot, that summarize how faithfully the distances between points on a multidimensional scaling map reflect the original dissimilarities.
  • Structural Equation Modeling (SEM)analysisAn analysis that defines concepts which cannot be measured directly through several observed variables and tests the hypothesised direct and indirect causal relations among those concepts together, in a single model.
  • Subgroup analysisconceptA meta-analytic method that splits studies into groups by a categorical characteristic (moderator), computes a separate pooled effect in each group and tests the difference between them.
  • Subsample ratio (stochastic boosting)conceptA setting in gradient boosting that makes each new tree be built on a random portion of the training data rather than all of it, and that sets the size of that portion.
  • Supplementary (passive) pointsconceptRow or column categories in correspondence analysis that do not take part in determining the solution (the axes) but whose profiles are placed on the map afterwards for comparison with the other categories.
  • Support Vector Machine (SVM)analysisA classification and regression method that finds the boundary separating two classes with the widest possible gap (margin), and can build curved boundaries through kernel functions when needed.
  • Survey-PHREG (Weighted Cox)analysisA method for survival data collected with a weighted, stratified and clustered sampling design (e.g. a national health survey) that estimates Cox regression so as to represent the whole population and computes standard errors according to the sampling design.
  • Survival confidence interval transformations (log-log)conceptScale transformations used when building confidence intervals for the Kaplan-Meier survival probability S(t), so that the interval stays within 0-1 and achieves better coverage.
  • SVM kernel function (rbf, linear, polynomial) and γconceptThe function in a support vector machine that defines the similarity between observations and so determines the shape of the decision boundary (straight, curved, flexible); in kernels such as rbf, γ sets how local this similarity is.
  • Šidák correctionconceptA multiple-comparison correction that lowers the significance level of each test to 1 − (1 − α)^(1/m) so that the probability of at least one false positive across all tests stays at α.
T
  • t-SNE (t-Distributed Stochastic Neighbor Embedding)analysisA non-linear visualisation method that places observations described by many variables onto a two- or three-dimensional map so that those close to one another stay close.
  • Taylor linearization and replication variance methodsconceptThe two main ways of obtaining standard errors of weighted estimates in complex sample designs while accounting for stratification and clustering.
  • Ties and zero differences (rank tests)conceptIn rank-based tests, observations sharing the same value (ties) and, in paired data, pairs whose difference is exactly zero; both require special handling of the test statistic.
  • Ties method (Efron, Breslow, exact)conceptThe method that determines how the partial likelihood of Cox regression is computed when several events occur at the same time (tied times).
  • Time SeriesanalysisA descriptive and testing examination of a single time series before modelling: whether it is stationary, how strongly it depends on its own past and whether it contains seasonality.
  • Time-dependent covariate and counting-process dataconceptA predictor whose value changes during follow-up, and the data layout that splits each person's follow-up into (start, stop] intervals so that such predictors can enter survival models.
  • Time-dependent Cox RegressionanalysisA Cox regression for situations in which a person's characteristics change during follow-up (e.g. starting treatment later) or a factor's effect changes over time, modelling the risk of the event at each moment from the values at that moment.
  • Tobit Regression (Censored Regression)analysisA regression model for outcomes that pile up at a lower or upper limit (e.g. many zeros), which treats values beyond the limit as censored when estimating the effects of the predictors.
  • Train-test split (holdout)conceptSetting aside part of the data to build the model (training) and the rest to measure how well it performs on data it has not seen (testing).
  • Tree pruning and the complexity parameter (cp)conceptThe process of turning an overgrown decision tree into a simpler, more generalizable one by removing branches that do not add enough to predictive performance; cp sets how far this simplification goes.
  • Trim-and-fillconceptA meta-analysis method that uses funnel-plot asymmetry to estimate how many studies may be missing and gives an adjusted pooled effect after adding them.
  • Trimmed meanconceptThe arithmetic mean of the values that remain after a fixed proportion of the smallest and largest observations has been removed from each end of the data.
  • True Skill Statistic (TSS)analysisA measure that summarises how skilfully a classifier or species distribution model separates presence from absence at a given threshold, using the formula sensitivity + specificity − 1.
  • TruncationconceptA situation in which units beyond a certain limit never enter the sample, so neither their values nor their existence is observed.
  • Trustworthiness and continuity measuresconceptTwo rank-based measures of how well a low-dimensional embedding from methods such as t-SNE and UMAP preserves the local neighbourhoods of the original space.
  • Tukey's HSD testconceptA multiple-comparison test that, after a significant ANOVA, compares the means of every pair of groups while keeping the familywise Type I error rate at α.
  • Tukey's test for nonadditivityconceptA one-degree-of-freedom test for a two-way layout with a single observation per cell (e.g. persons × items) that checks whether the row and column effects are additive, i.e. whether there is a multiplicative interaction between them.
  • Turnbull NPMLE (interval-censored data)conceptThe nonparametric maximum likelihood estimate of the survival function for data in which the event time is not known exactly, only that it fell in an interval between two examinations.
  • Two-stage least squares (2SLS) and LIMLconceptTwo methods that estimate the effect of an explanatory variable consistently, with the help of instrumental variables, when that variable is correlated with the error term (endogenous).
  • Two-Way ANOVAanalysisAn analysis of variance that tests, in one model, the separate effects of two factors on a continuous outcome (main effects) and how they work together (interaction).
  • Two-way fixed effects (TWFE)conceptA regression model for panel data that controls, through unit and time fixed effects, both for each unit's time-invariant characteristics and for period shocks that affect all units at once.
  • Type I and Type II errorsconceptIn a hypothesis test, rejecting a true null hypothesis is a Type I error, and failing to reject a false null hypothesis is a Type II error.
  • Types of sums of squares (Type I, II, III)conceptOptions that set, in unbalanced designs, which other effects in the model an effect's sum of squares is adjusted for.
U
  • UMAP (Uniform Manifold Approximation and Projection)analysisA fast, non-linear dimension reduction method based on a neighbourhood graph, which places multivariate observations onto a map of two or a few dimensions while preserving each one's near neighbours.
  • Unconditional exact tests (Barnard, Boschloo)conceptExact tests for comparing two independent proportions in a 2×2 table that fix only the group sizes and are usually more powerful than Fisher's test.
  • UnidimensionalityconceptThe assumption that responses to the items of a scale are driven essentially by a single latent trait (an ability or attitude).
  • Unit root tests (ADF, Phillips-Perron)conceptTests of whether a time series contains a unit root, i.e. whether it is a non-stationary process (e.g. a random walk) in which shocks have permanent effects.
  • Unobserved Components Model (Structural Time Series)analysisA structural time series model that writes a series as the sum of components that cannot be observed directly (level, slope, season, cycle and irregular) and estimates them with the Kalman filter to produce forecasts.
V
  • VAR (Vector Autoregression)analysisA vector autoregression method that models several mutually influencing time series together, with a system of equations that explains each series by its own past and by the past of the other series.
  • Variable Clustering (VARCLUS)analysisA method that sorts highly correlated numeric variables into groups, represents each group by a single summary component and shows the best representative variable of each group.
  • Variable importance (gain, cover, frequency)conceptMeasures in tree-based and other machine-learning models that summarize how much each predictor contributes to the model's predictive performance.
  • Variance componentsconceptThe shares into which the total variability of a measurement is divided among different sources modelled as random effects (e.g. subjects, raters, groups, error).
  • Variance Components Analysis (VARCOMP)analysisAn analysis that splits the total variability in a measurement into sources treated as random effects (such as batches, operators, schools, animals) and measurement error, estimating the share of each source.
  • Variogram (semivariogram)conceptA function describing how dissimilarity between two points in spatial data changes with the distance between them; it is the basis of kriging interpolation.
  • Variogram + Kriging (Spatial Interpolation)analysisA spatial interpolation method that models how similarity between measurement points declines with distance using a variogram, and uses this model to estimate values, and the uncertainty of those estimates, at unmeasured locations.
  • Vector error correction model (VECM)conceptA model for non-stationary time series that share a long-run equilibrium relationship (cointegration), capturing both their short-run fluctuations and their return toward equilibrium.
W
  • Wald test and Wald confidence intervalconceptA test that asks how many standard errors a parameter estimate lies from its null value, and the confidence interval of the form 'estimate ± z × standard error' built on the same idea.
  • Weighted Frequency (Survey Freq)analysisAn analysis that estimates the population proportions and counts of a categorical variable, with design-correct confidence intervals, from survey data collected under a complex sampling design, and tests the association of two categorical variables while accounting for the design.
  • Weighted kappaconceptA version of Cohen's kappa for ordered categories that credits not only exact agreement between two raters but also partial agreement in neighbouring categories.
  • Weighted least squares (WLS)conceptA regression estimation method that gives each observation its own weight and minimises the weighted sum of squared residuals; it is mainly used when observations have different variances.
  • Weighted Linear Regression (Survey Regression)analysisA linear regression that relates a continuous outcome to predictors in survey data collected under a complex sampling design, estimating the coefficients with sampling weights and the standard errors in a design-correct way.
  • Weighted Logistic Regression (Survey Logistic Regression)analysisA logistic regression that relates the probability of a yes/no outcome to predictors in survey data collected under a complex sampling design, estimating the coefficients with sampling weights and the standard errors in a design-correct way.
  • Weighted Mean (Survey Means)analysisAn analysis that estimates the population mean of a variable, together with a design-correct standard error, from survey data collected under a complex sampling design (weights, strata, clusters).
  • Weighted Total (Survey Total)analysisAn analysis that estimates the population total of a variable (such as a total number of people, total area or total spending), together with its design-correct standard error, from sample data collected under a complex sampling design.
  • Welch correctionconceptA method that corrects the standard error and degrees of freedom of the t-test and ANOVA when the group variances are not equal.
  • White noiseconceptA time series with a constant mean (usually zero), constant variance and values at different times that are uncorrelated with each other; pure randomness with no predictable structure.
  • Wilcoxon Signed-RankanalysisA test that needs no normality: it checks whether the differences between two measurements on the same people are balanced around zero or shift systematically in one direction, using the ranks of the differences' sizes together with their signs.
  • Within-subject (repeated-measures) factorconceptA factor whose levels are all measured on the same subjects, such as measuring the same people at different times or under different conditions.
Y
  • Youden's J index (optimal cut-off)conceptA measure, J = sensitivity + specificity - 1, that combines a diagnostic test's or classifier's sensitivity and specificity in one number, and the cut-off that maximizes it.
Z
  • z-score (standardization)conceptA unit-free score showing how many standard deviations a value lies from the mean: z = (x - mean) / standard deviation.
  • Zero-inflated and hurdle modelsconceptTwo families of models for count data with far more zeros than a standard Poisson or negative binomial model predicts, which explain the zeros with a separate process.