Data with categorical values that can be ordered by magnitude, but the exact distance (spacing) between categories is undefined or unknown.
Ordinal data is one of the levels in Stevens' theory of scale types, which also includes nominal, interval and ratio data. This scheme can be useful and is widely applied, but is not without its problems, and some prominent statisticians think it is misleading.
Use ordered-logit for ordinal logistic regression, which is a type of regression that is applicable when the dependent variable is ordinal. Newer methods of regression may also apply penalization to reduce overfitting in linear models with ordinal predictors.