Practical, careful guidance for planning, measurement, modeling, validation, and reproducible quantitative work.
Search by question, method, model, or research topic.
79 decision guides
A synthetic case study showing why ordinary regression can be misleading when daily observations are serially dependent, and how trend, weekly structure, AR(1) disturbances, residual diagnostics, and forecast uncertainty change the analysis.
Read guideFour of five endpoints have p < .05. Does that mean a treatment worked on four outcomes? This synthetic clinical-trial case shows how multiplicity adjustment, family-wise error control, hypothesis hierarchy, and gatekeeping can lead to very different—but methodologically coherent—conclusions.
Read guideA high development-sample AUC does not by itself validate a clinical prediction model. This Resource explains how discrimination, calibration, optimism-corrected internal validation, external validation, updating, and clinical usefulness contribute to a defensible validation strategy.
Read guideLearn how to interpret funnel plot asymmetry, symmetry, and small-study effects without treating them as definitive evidence for or against publication bias. This Resource explains how publication-bias procedures and adjustments should be used to assess robustness rather than reconstruct an unknowable “true” effect.
Read guideKaplan–Meier, the log-rank test, and Cox regression all analyze time-to-event data, but they answer different research questions. This Resource explains how to choose among description, unadjusted group comparison, adjusted hazard modeling, and prediction while accounting for censoring, truncation, proportional hazards, and correct hazard-ratio interpretation.
Read guideMean-centering, standardization, and changing predictor units affect regression coefficients in different ways. This Resource explains what each transformation changes, how interpretation shifts in interaction models, and why centering or standardizing should not be applied mechanically.
Read guideRepeated measurements alone do not make standard regression inappropriate. This Resource explains when treatment-confounder feedback creates a more difficult longitudinal causal problem, why conventional adjustment can fail, and when methods such as inverse probability weighting and marginal structural models may be appropriate.
Read guideA practical checklist for assessing whether a fitted regression model is adequately specified, diagnosed, and—when prediction is the goal—validated. It covers data structure, residuals, influence, functional form, fit, uncertainty, logistic regression diagnostics, and predictive validation.
Read guideLearn how study design determines whether observations are paired or independent, and how that dependence structure changes the appropriate statistical analysis for continuous, rankable, binary, categorical, and diagnostic-test outcomes.
Read guideSensitivity, specificity, PPV, and NPV answer different conditional-probability questions, while ROC curves and AUC address threshold-dependent discrimination. This Resource provides a practical framework for interpreting these measures without confusing predictive values, prevalence effects, likelihood ratios, ROC performance, and clinical threshold selection.
Read guideMeasurement error and misclassification can systematically distort associations, causal effects, confounding control, and diagnostic accuracy even in large, precisely analyzed studies. This Resource provides a practical framework for recognizing measurement problems, evaluating their likely consequences, using validation or calibration information, and interpreting results appropriately.
Read guideLearn how to interpret a nonsignificant result using effect estimates, confidence intervals, precision, and sample size—and why a p value greater than .05 does not by itself demonstrate that there is no effect.
Read guideYour design, measurement, data structure, assumptions, and existing output can change the appropriate statistical decision. If the choice is still unclear, discuss the study in context.