Rank-based methods are non-parametric statistical techniques that rely on the order or rank of data rather than their specific values, making them robust against outliers and distributional assumptions. These methods are particularly useful for analyzing ordinal data or when the assumptions of parametric tests, such as normality, are violated.