Modeling Outcomes With Floor or Ceiling Effects: An Introduction to the Tobit Model

In gifted education research, it is common for outcome variables to exhibit strong floor or ceiling effects due to insufficient range of measurement of many instruments when used with gifted populations. Common statistical methods (e.g., analysis of variance, linear regression) produce biased estimates when such effects are present. In practice, it is frequent for researchers to ignore ceiling effects and proceed with traditional analysis. However, the problems caused by ceiling effects are not without possible solutions. This Methodological Brief describes a variation of multiple regression, called the Tobit model, which is capable of correct inference when floor or ceiling effects are present. A brief simulation study illustrates the performance of the Tobit model with a dataset exhibiting a ceiling effect.