Multivariable vs Multivariate

These two words are constantly mixed up, and getting them right is a quick marker of statistical literacy. Multivariable analysis means a model with ONE outcome and MULTIPLE predictors, exactly the adjusted linear, logistic, and Cox models from this module. When a paper says it 'adjusted for confounders' or reports 'adjusted odds ratios', it is doing multivariable analysis. This is the workhorse of clinical research. Multivariate analysis, strictly speaking, means analyzing MULTIPLE outcomes (dependent variables) at the same time, for example MANOVA (multivariate analysis of variance), or models that jointly analyze several correlated outcomes. These are far less common in everyday clinical papers. In practice, many authors and journals loosely call any model with several predictors 'multivariate', which is technically incorrect; the precise term is 'multivariable'. You do not need to be pedantic in conversation, but knowing the distinction helps you read methods sections accurately and describe your own analyses correctly. Why does it matter? The adjustment that lets regression isolate one predictor's effect while holding others constant, the entire point of controlling for confounders, is a multivariable technique, and it gives observational research much of its credibility. When you see 'adjusted' estimates, you are seeing multivariable modeling at work; reserve 'multivariate' for the genuinely multi-outcome situation. A simple memory aid: multivariABLE means many variABLES (predictors) with one outcome, while multivariATE means many outcomes. When you read a methods section, do not be thrown by the label: the question to ask is how many outcomes are being modeled. One outcome, even with twenty predictors, is multivariable; only when several dependent variables are analyzed together is the analysis truly multivariate. The distinction rarely changes a study's validity, but using the right word, and recognizing the very common misuse, marks careful statistical reporting and helps you describe your own analysis correctly to reviewers and readers.