Practical, careful guidance for planning, measurement, modeling, validation, and reproducible quantitative work.
Search by question, method, model, or research topic.
79 decision guides
Stepwise regression and univariable screening use observed data to determine the final model, adding selection uncertainty and potential overfitting. This Resource explains why that matters and how resampling, shrinkage, ridge, and lasso can support more defensible model development and validation.
Read guideCategorizing continuous variables can discard information, reduce precision, create artificial boundaries, and weaken regression and prediction models. This Resource explains when to keep a predictor continuous, when to consider flexible modeling, and when a scientifically established threshold may justify categorization.
Read guideThere is no universal SEM minimum that makes 200 participants automatically adequate or inadequate. This synthetic case shows how latent variables, indicators, model identification, measurement quality, structural effects, estimation, missingness, and model-specific power change the sample-size decision.
Read guideLearn why repeated testing creates more opportunities for false-positive findings and how to define the inferential family, distinguish confirmatory from exploratory analyses, choose an appropriate error-control strategy, and report multiple comparisons transparently.
Read guideStudents from the same school share context, so treating every student as an independent observation can overstate statistical precision. This synthetic case study shows how multilevel modeling separates student- and school-level predictors, calculates the ICC, and uses random intercepts and random slopes to represent nested education data.
Read guideUnderstand how Bayesian and frequentist statistics differ in medical research, including confidence and credible intervals, priors, posterior inference, prediction, hierarchical models, model checking, and clinical trials. The Resource focuses on matching the inferential framework to the actual scientific or decision problem.
Read guideThis synthetic case study shows why time-to-event outcomes with right censoring require survival methods rather than simple comparisons of mean follow-up or event status. It demonstrates Kaplan–Meier estimation, the log-rank test, Cox proportional hazards regression, and careful interpretation of hazard ratios.
Read guideReflective and formative constructs represent different relationships between indicators and constructs. This synthetic PLS-SEM case study shows how construct conceptualization, indicator direction, interchangeability, collinearity, outer weights, loadings, reliability, and validity criteria lead to different measurement decisions.
Read guideLearn how categorical predictors should be represented in regression using dummy coding, why a four-category predictor typically requires three indicators, and how the reference category determines the interpretation of the intercept and coefficients.
Read guideStrong PLS-SEM path coefficients do not rescue weak measurement. This synthetic case study follows the model-assessment sequence from specification and data examination through reflective measurement quality, validity, collinearity, bootstrapping, structural interpretation, and prediction.
Read guideA practical framework for interpreting meta-regression and subgroup analysis without overstating study-level moderator findings. It explains how prespecification, multiplicity, statistical power, effect scale, ecological interpretation, and residual heterogeneity should shape conclusions.
Read guideExternal validation can reveal changed calibration without implying that the original prediction model should be discarded. This Resource explains how to choose the least extensive update that adequately addresses the observed transport problem, from baseline-risk recalibration through coefficient revision, model extension, and redevelopment.
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.